v3.16.4

RELEASE_NOTES

@@ -65,6 +65,11 @@ If not what makes it happen or not happen?
 Change Log
 ----------
 
+v3.16.4
+
+Faster sha512 and sha256 when not using SHA CPU extension.
+#329: Fixed GBT incorrect target diff in stats.
+
 v3.16.3
 
 #313 Fix compile error with GCC 11.

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

@@ -74,9 +74,15 @@ static const uint32_t K256[64] =
 #define CHs(X, Y, Z) \
    _mm_xor_si128( _mm_and_si128( _mm_xor_si128( Y, Z ), X ), Z ) 
 
+/*
 #define MAJs(X, Y, Z) \
    _mm_or_si128( _mm_and_si128( X, Y ), \
                     _mm_and_si128( _mm_or_si128( X, Y ), Z ) )
+*/
+
+#define MAJs(X, Y, Z) \
+  _mm_xor_si128( Y, _mm_and_si128( _mm_xor_si128( X, Y ), \
+                                   _mm_xor_si128( Y, Z ) ) )
 
 #define BSG2_0(x) \
    _mm_xor_si128( _mm_xor_si128( \
@@ -345,9 +351,20 @@ void sha256_4way_full( void *dst, const void *data, size_t len )
 #define CHx(X, Y, Z) \
    _mm256_xor_si256( _mm256_and_si256( _mm256_xor_si256( Y, Z ), X ), Z ) 
 
+/*
 #define MAJx(X, Y, Z) \
    _mm256_or_si256( _mm256_and_si256( X, Y ), \
                     _mm256_and_si256( _mm256_or_si256( X, Y ), Z ) )
+*/
+/*
+#define MAJx(X, Y, Z) \
+  _mm256_xor_si256( Y, _mm256_and_si256( _mm256_xor_si256( X, Y ), \
+                                         _mm256_xor_si256( Y, Z ) ) )
+*/
+
+#define MAJx(X, Y, Z) \
+  _mm256_xor_si256( Y, _mm256_and_si256( X_xor_Y = _mm256_xor_si256( X, Y ), \
+                                         Y_xor_Z ) )
 
 #define BSG2_0x(x) \
    _mm256_xor_si256( _mm256_xor_si256( \
@@ -375,14 +392,15 @@ do { \
   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) ); \
+  Y_xor_Z = X_xor_Y; \
   D  = _mm256_add_epi32( D,  T1 ); \
   H  = _mm256_add_epi32( T1, T2 ); \
 } while (0)
 
 static void
 sha256_8way_round( sha256_8way_context *ctx,  __m256i *in, __m256i r[8] )
 {
-   register  __m256i A, B, C, D, E, F, G, H;
+   register  __m256i A, B, C, D, E, F, G, H, X_xor_Y, Y_xor_Z;
    __m256i W[16];
 
    mm256_block_bswap_32( W  , in   );
@@ -411,6 +429,8 @@ sha256_8way_round( sha256_8way_context *ctx,  __m256i *in, __m256i r[8] )
       H = m256_const1_64( 0x5BE0CD195BE0CD19 );
    }
 
+   Y_xor_Z = _mm256_xor_si256( B, C );
+
    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 );
    SHA2s_8WAY_STEP( G, H, A, B, C, D, E, F,  2, 0 );
@@ -591,9 +611,20 @@ void sha256_8way_full( void *dst, const void *data, size_t len )
 #define CHx16(X, Y, Z) \
    _mm512_xor_si512( _mm512_and_si512( _mm512_xor_si512( Y, Z ), X ), Z ) 
 
+/*
 #define MAJx16(X, Y, Z) \
    _mm512_or_si512( _mm512_and_si512( X, Y ), \
                     _mm512_and_si512( _mm512_or_si512( X, Y ), Z ) )
+*/
+/*
+#define MAJx16(X, Y, Z) \
+  _mm512_xor_si512( Y, _mm512_and_si512( _mm512_xor_si512( X, Y ), \
+                                         _mm512_xor_si512( Y, Z ) ) )
+*/
+
+#define MAJx16(X, Y, Z) \
+  _mm512_xor_si512( Y, _mm512_and_si512( X_xor_Y = _mm512_xor_si512( X, Y ), \
+                                         Y_xor_Z ) )
 
 #define BSG2_0x16(x) \
    _mm512_xor_si512( _mm512_xor_si512( \
@@ -621,14 +652,15 @@ do { \
   T1 = _mm512_add_epi32( H, mm512_add4_32( BSG2_1x16(E), CHx16(E, F, G), \
                                            K, W[i] ) ); \
   T2 = _mm512_add_epi32( BSG2_0x16(A), MAJx16(A, B, C) ); \
+  Y_xor_Z = X_xor_Y; \
   D  = _mm512_add_epi32( D,  T1 ); \
   H  = _mm512_add_epi32( T1, T2 ); \
 } while (0)
 
 static void
 sha256_16way_round( sha256_16way_context *ctx,  __m512i *in, __m512i r[8] )
 {
-   register  __m512i A, B, C, D, E, F, G, H;
+   register  __m512i A, B, C, D, E, F, G, H, X_xor_Y, Y_xor_Z;
    __m512i W[16];
 
    mm512_block_bswap_32( W  , in   );
@@ -657,6 +689,8 @@ sha256_16way_round( sha256_16way_context *ctx,  __m512i *in, __m512i r[8] )
       H = m512_const1_64( 0x5BE0CD195BE0CD19 );
    }
 
+   Y_xor_Z = _mm512_xor_si512( B, C );
+
    SHA2s_16WAY_STEP( A, B, C, D, E, F, G, H,  0, 0 );
    SHA2s_16WAY_STEP( H, A, B, C, D, E, F, G,  1, 0 );
    SHA2s_16WAY_STEP( G, H, A, B, C, D, E, F,  2, 0 );

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

@@ -98,9 +98,21 @@ static const uint64_t K512[80] =
 #define CH8W(X, Y, Z) \
    _mm512_xor_si512( _mm512_and_si512( _mm512_xor_si512( Y, Z ), X ), Z ) 
 
+/*
 #define MAJ8W(X, Y, Z) \
    _mm512_or_si512( _mm512_and_si512( X, Y ), \
                     _mm512_and_si512( _mm512_or_si512( X, Y ), Z ) )
+*/
+/* Functionally identical to original but optimizable,
+ * subexpression X^Y from one step can be reused in the next step as Y^Z
+#define MAJ8W(X, Y, Z) \
+  _mm512_xor_si512( Y, _mm512_and_si512( _mm512_xor_si512( X, Y ), \
+                                         _mm512_xor_si512( Y, Z ) ) )
+*/
+
+#define MAJ8W(X, Y, Z) \
+  _mm512_xor_si512( Y, _mm512_and_si512( X_xor_Y = _mm512_xor_si512( X, Y ), \
+                                         Y_xor_Z ) )
 
 #define BSG8W_5_0(x) \
    _mm512_xor_si512( _mm512_xor_si512( \
@@ -172,6 +184,7 @@ do { \
   T1 = _mm512_add_epi64( H, mm512_add4_64( BSG8W_5_1(E), CH8W(E, F, G), \
                                            K, W[i] ) ); \
   T2 = _mm512_add_epi64( BSG8W_5_0(A), MAJ8W(A, B, C) ); \
+  Y_xor_Z = X_xor_Y; \
   D  = _mm512_add_epi64( D, T1 ); \
   H  = _mm512_add_epi64( T1, T2 ); \
 } while (0)
@@ -180,7 +193,7 @@ static void
 sha512_8way_round( sha512_8way_context *ctx,  __m512i *in, __m512i r[8] )
 {
    int i;
-   register __m512i A, B, C, D, E, F, G, H;
+   register __m512i A, B, C, D, E, F, G, H, X_xor_Y, Y_xor_Z;
    __m512i W[80];
 
    mm512_block_bswap_64( W  , in );
@@ -213,6 +226,8 @@ sha512_8way_round( sha512_8way_context *ctx,  __m512i *in, __m512i r[8] )
       H = m512_const1_64( 0x5BE0CD19137E2179 );
    }
 
+   Y_xor_Z = _mm512_xor_si512( B, C );
+
    for ( i = 0; i < 80; i += 8 )
    {
       SHA3_8WAY_STEP( A, B, C, D, E, F, G, H, i + 0 );
@@ -319,22 +334,28 @@ void sha512_8way_close( sha512_8way_context *sc, void *dst )
 
 // SHA-512 4 way 64 bit
 
-/*
+
 #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 MAJ(X, Y, Z) \
+  _mm256_xor_si256( Y, _mm256_and_si256( X_xor_Y = _mm256_xor_si256( X, Y ), \
+                                         Y_xor_Z ) )
+
 #define BSG5_0(x) \
   mm256_ror_64( _mm256_xor_si256( mm256_ror_64( \
                    _mm256_xor_si256( mm256_ror_64( x,  5 ), x ), 6 ), x ), 28 )
 
 #define BSG5_1(x) \
   mm256_ror_64( _mm256_xor_si256( mm256_ror_64( \
                    _mm256_xor_si256( mm256_ror_64( x, 23 ), x ), 4 ), x ), 14 )
-*/
+
 /*
 #define BSG5_0(x) \
    _mm256_xor_si256( _mm256_xor_si256( \
@@ -402,7 +423,7 @@ static inline __m256i ssg512_add( __m256i w0, __m256i w1 )
   w1  = _mm256_xor_si256( X1a, X1b ); \
 } while(0)
 */
-
+/*
 #define SHA3_4WAY_STEP(A, B, C, D, E, F, G, H, i) \
 do { \
   __m256i K = _mm256_set1_epi64x( K512[ i ] ); \
@@ -431,7 +452,7 @@ do { \
   H  = _mm256_add_epi64( T1, T2 ); \
   D  = _mm256_add_epi64( D, T1 ); \
 } while (0)
-
+*/
 /*
 #define SHA3_4WAY_STEP(A, B, C, D, E, F, G, H, i) \
 do { \
@@ -445,24 +466,25 @@ do { \
 } 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) ); \
+  Y_xor_Z = X_xor_Y; \
   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;
+   register __m256i A, B, C, D, E, F, G, H, X_xor_Y, Y_xor_Z;
    __m256i W[80];
 
    mm256_block_bswap_64( W  , in );
@@ -495,6 +517,8 @@ sha512_4way_round( sha512_4way_context *ctx,  __m256i *in, __m256i r[8] )
       H = m256_const1_64( 0x5BE0CD19137E2179 );
    }
 
+   Y_xor_Z = _mm256_xor_si256( B, C );
+
    for ( i = 0; i < 80; i += 8 )
    {
       SHA3_4WAY_STEP( A, B, C, D, E, F, G, H, i + 0 );

algo/sha/sph_sha2.c

@@ -40,8 +40,8 @@
 #endif
 
 #define CH(X, Y, Z)    ((((Y) ^ (Z)) & (X)) ^ (Z))
-#define MAJ(X, Y, Z)   (((Y) & (Z)) | (((Y) | (Z)) & (X)))
-
+//#define MAJ(X, Y, Z)   (((Y) & (Z)) | (((Y) | (Z)) & (X)))
+#define MAJ( X, Y, Z )   ( Y  ^ ( ( X ^ Y ) & ( Y ^ Z ) ) )
 #define ROTR    SPH_ROTR32
 
 #define BSG2_0(x)      (ROTR(x, 2) ^ ROTR(x, 13) ^ ROTR(x, 22))

algo/sha/sph_sha2big.c

@@ -38,7 +38,8 @@
 #if SPH_64
 
 #define CH(X, Y, Z)    ((((Y) ^ (Z)) & (X)) ^ (Z))
-#define MAJ(X, Y, Z)   (((X) & (Y)) | (((X) | (Y)) & (Z)))
+//#define MAJ(X, Y, Z)   (((X) & (Y)) | (((X) | (Y)) & (Z)))
+#define MAJ( X, Y, Z )   ( Y  ^ ( ( X ^ Y ) & ( Y ^ Z ) ) )
 
 #define ROTR64    SPH_ROTR64
 

configure

@@ -1,6 +1,6 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.69 for cpuminer-opt 3.16.3.
+# Generated by GNU Autoconf 2.69 for cpuminer-opt 3.16.4.
 #
 #
 # Copyright (C) 1992-1996, 1998-2012 Free Software Foundation, Inc.
@@ -577,8 +577,8 @@ MAKEFLAGS=
 # Identity of this package.
 PACKAGE_NAME='cpuminer-opt'
 PACKAGE_TARNAME='cpuminer-opt'
-PACKAGE_VERSION='3.16.3'
-PACKAGE_STRING='cpuminer-opt 3.16.3'
+PACKAGE_VERSION='3.16.4'
+PACKAGE_STRING='cpuminer-opt 3.16.4'
 PACKAGE_BUGREPORT=''
 PACKAGE_URL=''
 
@@ -1332,7 +1332,7 @@ if test "$ac_init_help" = "long"; then
   # Omit some internal or obsolete options to make the list less imposing.
   # This message is too long to be a string in the A/UX 3.1 sh.
   cat <<_ACEOF
-\`configure' configures cpuminer-opt 3.16.3 to adapt to many kinds of systems.
+\`configure' configures cpuminer-opt 3.16.4 to adapt to many kinds of systems.
 
 Usage: $0 [OPTION]... [VAR=VALUE]...
 
@@ -1404,7 +1404,7 @@ fi
 
 if test -n "$ac_init_help"; then
   case $ac_init_help in
-     short | recursive ) echo "Configuration of cpuminer-opt 3.16.3:";;
+     short | recursive ) echo "Configuration of cpuminer-opt 3.16.4:";;
    esac
   cat <<\_ACEOF
 
@@ -1509,7 +1509,7 @@ fi
 test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
-cpuminer-opt configure 3.16.3
+cpuminer-opt configure 3.16.4
 generated by GNU Autoconf 2.69
 
 Copyright (C) 2012 Free Software Foundation, Inc.
@@ -2012,7 +2012,7 @@ cat >config.log <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
-It was created by cpuminer-opt $as_me 3.16.3, which was
+It was created by cpuminer-opt $as_me 3.16.4, which was
 generated by GNU Autoconf 2.69.  Invocation command line was
 
   $ $0 $@
@@ -2993,7 +2993,7 @@ fi
 
 # Define the identity of the package.
  PACKAGE='cpuminer-opt'
- VERSION='3.16.3'
+ VERSION='3.16.4'
 
 
 cat >>confdefs.h <<_ACEOF
@@ -6690,7 +6690,7 @@ cat >>$CONFIG_STATUS <<\_ACEOF || ac_write_fail=1
 # report actual input values of CONFIG_FILES etc. instead of their
 # values after options handling.
 ac_log="
-This file was extended by cpuminer-opt $as_me 3.16.3, which was
+This file was extended by cpuminer-opt $as_me 3.16.4, which was
 generated by GNU Autoconf 2.69.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
@@ -6756,7 +6756,7 @@ _ACEOF
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 ac_cs_config="`$as_echo "$ac_configure_args" | sed 's/^ //; s/[\\""\`\$]/\\\\&/g'`"
 ac_cs_version="\\
-cpuminer-opt config.status 3.16.3
+cpuminer-opt config.status 3.16.4
 configured by $0, generated by GNU Autoconf 2.69,
   with options \\"\$ac_cs_config\\"
 

configure.ac

@@ -1,4 +1,4 @@
-AC_INIT([cpuminer-opt], [3.16.3])
+AC_INIT([cpuminer-opt], [3.16.4])
 
 AC_PREREQ([2.59c])
 AC_CANONICAL_SYSTEM

cpu-miner.c

@@ -447,8 +447,10 @@ static bool work_decode( const json_t *val, struct work *work )
 
     if ( !allow_mininginfo )
         net_diff = algo_gate.calc_network_diff( work );
+    else
+        net_diff = hash_to_diff( work->target );
 
-    work->targetdiff = hash_to_diff( work->target );
+    work->targetdiff = net_diff;
     stratum_diff = last_targetdiff = work->targetdiff;
     work->sharediff = 0;
     algo_gate.decode_extra_data( work, &net_blocks );
@@ -908,7 +910,8 @@ static bool gbt_work_decode( const json_t *val, struct work *work )
    }
    for ( i = 0; i < ARRAY_SIZE( work->target ); i++ )
       work->target[7 - i] = be32dec( target + i );
-
+   net_diff = work->targetdiff = hash_to_diff( work->target );
+
    tmp = json_object_get( val, "workid" );
    if ( tmp )
    {