v25.2

Makefile.am

@@ -1,19 +1,39 @@
 
-if WANT_JANSSON
-JANSSON_INCLUDES= -I$(top_srcdir)/compat/jansson
+if HAVE_APPLE
+# MacOS uses Homebrew to install needed packages but they aren't linked for
+# the jansson test in configure. Ignore the failed test & link them now,
+# different path for different CPU arch.
+
+if ARCH_ARM64
+  EXTRA_INCLUDES = -I/opt/homebrew/include
+  EXTRA_LIBS     = -L/opt/homebrew/lib
 else
-JANSSON_INCLUDES=
+  EXTRA_INCLUDES = -I/usr/local/include
+  EXTRA_LIBS     = -L/usr/local/lib
 endif
 
-EXTRA_DIST	= example-cfg.json nomacro.pl
+else
 
-SUBDIRS		= compat
+if WANT_JANSSON
+# Can't find jansson libraries, compile the included source code.
+  EXTRA_INCLUDES = -I$(top_srcdir)/compat/jansson
+  EXTRA_LIBS     = -L$(top_srcdir)/compat/jansson
+else
+  EXTRA_INCLUDES =
+  EXTRA_LIBS     =
+endif
 
-ALL_INCLUDES	= @PTHREAD_FLAGS@ -fno-strict-aliasing $(JANSSON_INCLUDES) -I.
+endif
 
-bin_PROGRAMS	= cpuminer
+EXTRA_DIST = example-cfg.json nomacro.pl
 
-dist_man_MANS	= cpuminer.1
+SUBDIRS = compat
+
+ALL_INCLUDES = @PTHREAD_FLAGS@ -fno-strict-aliasing $(EXTRA_INCLUDES) -I.
+
+bin_PROGRAMS = cpuminer
+
+dist_man_MANS = cpuminer.1
 
 cpuminer_SOURCES = \
   dummy.cpp \
@@ -166,8 +186,6 @@ cpuminer_SOURCES = \
   algo/shavite/sph-shavite-aesni.c \
   algo/shavite/shavite-hash-2way.c \
   algo/shavite/shavite-hash-4way.c \
-  algo/simd/nist.c \
-  algo/simd/vector.c \
   algo/simd/sph_simd.c \
   algo/simd/simd-hash-2way.c \
   algo/skein/sph_skein.c \
@@ -274,25 +292,29 @@ cpuminer_SOURCES = \
   algo/yespower/yespower-opt.c \
   algo/yespower/yespower-ref.c \
   algo/yespower/yespower-blake2b-ref.c
-  
-disable_flags =
-
-if USE_ASM
-   cpuminer_SOURCES += asm/neoscrypt_asm.S
-else
-   disable_flags += -DNOASM
-endif
 
 if HAVE_WINDOWS
    cpuminer_SOURCES += compat/winansi.c
 endif
 
-cpuminer_LDFLAGS	= @LDFLAGS@
-cpuminer_LDADD	= @LIBCURL@ @JANSSON_LIBS@ @PTHREAD_LIBS@ @WS2_LIBS@ -lgmp
+if USE_ASM
+   disable_flags =
+   cpuminer_SOURCES += asm/neoscrypt_asm.S
+else
+   disable_flags += -DNOASM
+endif
+
+cpuminer_LDFLAGS = @LDFLAGS@
+cpuminer_LDADD	= $(EXTRA_LIBS) @LIBCURL@ -ljansson @PTHREAD_LIBS@ @WS2_LIBS@ -lgmp
 cpuminer_CPPFLAGS = @LIBCURL_CPPFLAGS@ $(ALL_INCLUDES)
-cpuminer_CFLAGS   = -Wno-pointer-sign -Wno-pointer-to-int-cast $(disable_flags)
+cpuminer_CFLAGS = -Wno-pointer-sign -Wno-pointer-to-int-cast $(disable_flags)
+
+if ARCH_ARM64
+   cpuminer_CFLAGS += -flax-vector-conversions
+endif
 
 if HAVE_WINDOWS
+
 # use to profile an object
 # gprof_cflags = -pg -g3
 # cpuminer_LDFLAGS += -pg
@@ -306,5 +328,4 @@ cpuminer-neoscrypt.o: neoscrypt.c
 	@echo "CUSTOM ${@}: ${filter %.o,${^}} ${filter %.c,${^}}"
 	$(CC) $(common_ccflags) -g -O3 $(gprof_cflags) -MT $@ -MD -MP -c -o $@ $<
 
-
 endif

README.md

@@ -36,34 +36,18 @@ for compile instructions.
 Requirements
 ------------
 
-1. A x86_64 architecture CPU with a minimum of SSE2 support. This includes
-Intel Core2 and newer and AMD equivalents. Further optimizations are available
-on some algoritms for CPUs with AES, AVX, AVX2, SHA, AVX512 and VAES.
-
-32 bit CPUs are not supported.
-Other CPU architectures such as ARM, Raspberry Pi, RISC-V, Xeon Phi, etc,
-are not supported.
+1. A 64 bit CPU supporting x86_64 (Intel or AMD) or aarch64 (ARM).
+x86_64 requires SSE2, aarch64 requires armv8 & NEON.
 
 Mobile CPUs like laptop computers are not recommended because they aren't
 designed for extreme heat of operating at full load for extended periods of
 time.
 
-Older CPUs and ARM architecture may be supported by cpuminer-multi by TPruvot.
-
-2. 64 bit Linux or Windows OS. Ubuntu and Fedora based distributions,
-including Mint and Centos, are known to work and have all dependencies
-in their repositories. Others may work but may require more effort. Older
-versions such as Centos 6 don't work due to missing features. 
-
-Windows 7 or newer is supported with mingw_w64 and msys or using the pre-built
-binaries. WindowsXP 64 bit is YMMV.
-
-FreeBSD is not actively tested but should work, YMMV.
-MacOS, OSx and Android are not supported.
+2. 64 bit operating system including Linux, Windows, MacOS, or BSD.
+Android, IOS and alt OSs like Haiku & ReactOS are not supported.
 
 3. Stratum pool supporting stratum+tcp:// or stratum+ssl:// protocols or
-RPC getwork using http:// or https://.
-GBT is YMMV.
+RPC getblockte,plate using http:// or https://.
 
 Supported Algorithms
 --------------------

RELEASE_NOTES

@@ -75,6 +75,31 @@ If not what makes it happen or not happen?
 Change Log
 ----------
 
+v25.2
+
+ARM: Fixed regression from v25.1 that could cause build fail.
+BSD: FreeBSD is now supported. Other BSDs may also work.
+MacOS: build with installed jansson library instead of compiling the included source code.
+Windows: remove "_WIN32_WINNT=0x0601" which is a downgrade on Win11.
+Changed build.sh shell from bash to sh.
+
+v25.1
+
+MacOS ARM64: m7m algo is now working.
+MacOS ARM64: can now be compiled with GCC.
+MacOS x86_64: is now working compiled with GCC.
+Fixed some minor bugs & removed some obsolete code.
+
+v24.8
+
+ARM: Apple MacOS on M series CPU is now supported compiled from source
+code, see Wiki for details.
+ARM: Fix incorrect compiler version display when using clang. 
+build.sh can now be used to compile all targets, arm_build.sh & build_msys2.sh
+have been removed.
+Windows: MSys2 build now enables CPU groups by default, prebuilt binaries
+continue to be compiled with CPU groups disabled.
+
 v24.7
 
 ARM: compile works for Windows using MSys2 & MingW, see wiki for details.

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

@@ -161,29 +161,25 @@ keccak64_8way_core( keccak64_ctx_m512i *kc, const void *data, size_t len,
 static void keccak64_8way_close( keccak64_ctx_m512i *kc, void *dst,
                                  size_t byte_len, size_t lim )
 {
-    unsigned eb;
-    union {
-       __m512i tmp[lim + 1];
-       uint64_t dummy;   /* for alignment */
-    } u;
+    __m512i tmp[lim + 1] __attribute__ ((aligned (64)));
     size_t j;
     size_t m512_len = byte_len >> 3;
+    const unsigned eb = hard_coded_eb;
 
-    eb = hard_coded_eb;
     if ( kc->ptr == (lim - 8) )
     {
         const uint64_t t = eb | 0x8000000000000000;
-        u.tmp[0] = _mm512_set1_epi64( t );
+        tmp[0] = _mm512_set1_epi64( t );
         j = 8;
     }
     else
     {
         j = lim - kc->ptr;
-        u.tmp[0] = _mm512_set1_epi64( eb );
-        memset_zero_512( u.tmp + 1, (j>>3) - 2 );
-        u.tmp[ (j>>3) - 1] = _mm512_set1_epi64( 0x8000000000000000 );
+        tmp[0] = _mm512_set1_epi64( eb );
+        memset_zero_512( tmp + 1, (j>>3) - 2 );
+        tmp[ (j>>3) - 1] = _mm512_set1_epi64( 0x8000000000000000 );
     }
-    keccak64_8way_core( kc, u.tmp, j, lim );
+    keccak64_8way_core( kc, tmp, j, lim );
     /* Finalize the "lane complement" */
     NOT64( kc->w[ 1], kc->w[ 1] );
     NOT64( kc->w[ 2], kc->w[ 2] );
@@ -361,29 +357,25 @@ keccak64_core( keccak64_ctx_m256i *kc, const void *data, size_t len,
 static void keccak64_close( keccak64_ctx_m256i *kc, void *dst, size_t byte_len,
             size_t lim )
 {
-    unsigned eb;
-    union {
-       __m256i tmp[lim + 1];
-       uint64_t dummy;   /* for alignment */
-    } u;
+    __m256i tmp[lim + 1] __attribute__ ((aligned (32)));
     size_t j;
     size_t m256_len = byte_len >> 3;
+    const unsigned eb = hard_coded_eb;
 
-    eb = hard_coded_eb;
     if ( kc->ptr == (lim - 8) )
     {
         const uint64_t t = eb | 0x8000000000000000;
-        u.tmp[0] = _mm256_set1_epi64x( t );
+        tmp[0] = _mm256_set1_epi64x( t );
         j = 8;
     }
     else
     {
         j = lim - kc->ptr;
-        u.tmp[0] = _mm256_set1_epi64x( eb );
-        memset_zero_256( u.tmp + 1, (j>>3) - 2 );
-        u.tmp[ (j>>3) - 1] = _mm256_set1_epi64x( 0x8000000000000000 );
+        tmp[0] = _mm256_set1_epi64x( eb );
+        memset_zero_256( tmp + 1, (j>>3) - 2 );
+        tmp[ (j>>3) - 1] = _mm256_set1_epi64x( 0x8000000000000000 );
     }
-    keccak64_core( kc, u.tmp, j, lim );
+    keccak64_core( kc, tmp, j, lim );
     /* Finalize the "lane complement" */
     NOT64( kc->w[ 1], kc->w[ 1] );
     NOT64( kc->w[ 2], kc->w[ 2] );

algo/m7m/m7m.c

@@ -1,8 +1,6 @@
 #include "cpuminer-config.h"
 #include "algo-gate-api.h"
 
-#if !defined(__APPLE__)
-
 #include <gmp.h>
 #include <stdbool.h>
 #include <stdlib.h>
@@ -33,6 +31,7 @@ static inline double exp_n( double xt )
         return exp( xt );
 }
 
+/*
 static inline double exp_n2( double x1, double x2 )
 {
     double p1 = -700., p2 = -37., p3 = -0.8e-8, p4 = 0.8e-8,
@@ -53,6 +52,7 @@ static inline double exp_n2( double x1, double x2 )
     else if ( xt > p6 - 1.e-200 )
         return 0.;
 }
+*/
 
 double swit2_( double wvnmb )
 {
@@ -298,14 +298,8 @@ int scanhash_m7m_hash( struct work* work, uint64_t max_nonce,
     return 0;
 }
 
-#endif   // not apple
-
 bool register_m7m_algo( algo_gate_t *gate )
 {
-#if defined(__APPLE__)
-  applog( LOG_ERR, "M7M algo is not supported on MacOS");
-  return false;
-#else  
   gate->optimizations = SHA256_OPT;
   init_m7m_ctx();
   gate->scanhash              = (void*)&scanhash_m7m_hash;
@@ -315,6 +309,5 @@ bool register_m7m_algo( algo_gate_t *gate )
   gate->set_work_data_endian  = (void*)&set_work_data_big_endian;
   opt_target_factor = 65536.0;
   return true;
-#endif
 }
 

algo/quark/hmq1725-4way.c

@@ -11,7 +11,6 @@
 #include "algo/luffa/luffa-hash-2way.h"
 #include "algo/cubehash/cube-hash-2way.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#include "algo/simd/nist.h"
 #include "algo/shavite/sph_shavite.h"
 #include "algo/shavite/shavite-hash-2way.h"
 #include "algo/simd/simd-hash-2way.h"
@@ -617,9 +616,9 @@ union _hmq1725_4way_context_overlay
     cubehashParam           cube;
     cube_2way_context       cube2;
     sph_shavite512_context  shavite;
-    hashState_sd            sd;
+    simd512_context         simd;
     shavite512_2way_context shavite2;
-    simd_2way_context       simd;
+    simd_2way_context       simd_2way;
     hashState_echo          echo;
     hamsi512_4way_context   hamsi;
     hashState_fugue         fugue;
@@ -753,8 +752,8 @@ extern void hmq1725_4way_hash(void *state, const void *input)
     shavite512_2way_full( &ctx.shavite2, vhashA, vhashA, 64 );
     shavite512_2way_full( &ctx.shavite2, vhashB, vhashB, 64 );
 
-    simd512_2way_full( &ctx.simd, vhashA, vhashA, 64 );
-    simd512_2way_full( &ctx.simd, vhashB, vhashB, 64 );
+    simd512_2way_full( &ctx.simd_2way, vhashA, vhashA, 64 );
+    simd512_2way_full( &ctx.simd_2way, vhashB, vhashB, 64 );
 
     rintrlv_2x128_4x64( vhash, vhashA, vhashB, 512 );     
 
@@ -869,41 +868,25 @@ extern void hmq1725_4way_hash(void *state, const void *input)
        echo_full( &ctx.echo, (BitSequence *)hash0, 512,
                        (const BitSequence *)hash0, 64 );
     else
-    {
-       init_sd( &ctx.sd, 512 );
-       update_final_sd( &ctx.sd, (BitSequence *)hash0,
-                           (const BitSequence *)hash0, 512 );
-    }
+       simd512_ctx( &ctx.simd, hash0, hash0, 64 );
 
    if ( hash1[0] & mask ) //4
        echo_full( &ctx.echo, (BitSequence *)hash1, 512,
                        (const BitSequence *)hash1, 64 );
    else
-   {
-       init_sd( &ctx.sd, 512 );
-       update_final_sd( &ctx.sd, (BitSequence *)hash1,
-                           (const BitSequence *)hash1, 512 );
-   }
+       simd512_ctx( &ctx.simd, hash1, hash1, 64 );
 
    if ( hash2[0] & mask ) //4
        echo_full( &ctx.echo, (BitSequence *)hash2, 512,
                        (const BitSequence *)hash2, 64 );
    else
-   {
-       init_sd( &ctx.sd, 512 );
-       update_final_sd( &ctx.sd, (BitSequence *)hash2,
-                           (const BitSequence *)hash2, 512 );
-   }
+       simd512_ctx( &ctx.simd, hash2, hash2, 64 );
 
    if ( hash3[0] & mask ) //4
        echo_full( &ctx.echo, (BitSequence *)hash3, 512,
                        (const BitSequence *)hash3, 64 );
    else
-   {
-       init_sd( &ctx.sd, 512 );
-       update_final_sd( &ctx.sd, (BitSequence *)hash3,
-                           (const BitSequence *)hash3, 512 );
-   }
+       simd512_ctx( &ctx.simd, hash3, hash3, 64 );
 
    intrlv_4x32( vhash, hash0, hash1, hash2, hash3, 512 );
 

algo/scrypt/neoscrypt.c

@@ -46,7 +46,7 @@
 #endif
 
 #ifdef __GNUC__
-#if defined(NOASM) || defined(__arm__) || defined(__aarch64__)
+#if defined(NOASM) || defined(__arm__) || defined(__aarch64__) || defined(__APPLE__)
 #define ASM 0
 #else
 #define ASM 1

algo/simd/nist.c

@@ -1,472 +0,0 @@
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-
-#include "nist.h"
-#include "simd_iv.h"
-
-
-/* #define NO_PRECOMPUTED_IV */
-#if defined(__SSE2__)  // || defined(__ARM_NEON)
-
-/* 
- * Increase the counter.
- */
-void IncreaseCounter(hashState_sd *state, DataLength databitlen) {
-#ifdef HAS_64
-      state->count += databitlen;
-#else
-      uint32_t old_count = state->count_low;
-      state->count_low += databitlen;
-      if (state->count_low < old_count)
-        state->count_high++;
-#endif
-}
-
-
-/* 
- * Initialize the hashState_sd with a given IV.
- * If the IV is NULL, initialize with zeros.
- */
-int InitIV(hashState_sd *state, int hashbitlen, const u32 *IV) {
-
-  int n = 8;
-
-  state->hashbitlen = hashbitlen;
-  state->n_feistels = n;
-  state->blocksize = 128*8;
-  
-#ifdef HAS_64
-  state->count = 0;
-#else
-  state->count_low  = 0;
-  state->count_high = 0;
-#endif  
-
-//  state->buffer = malloc(16*n + 16);
-  /*
-   * Align the buffer to a 128 bit boundary.
-   */
-//  state->buffer += ((unsigned char*)NULL - state->buffer)&15;
-
-//  state->A = malloc((4*n+4)*sizeof(u32));
-  /*
-   * Align the buffer to a 128 bit boundary.
-   */
-//  state->A += ((u32*)NULL - state->A)&3;
-
-  state->B = state->A+n;
-  state->C = state->B+n;
-  state->D = state->C+n;
-
-  if (IV)
-    memcpy(state->A, IV, 4*n*sizeof(u32));
-  else
-    memset(state->A, 0, 4*n*sizeof(u32));
-
-   // free(state->buffer);
-  //  free(state->A);	
-  return 0;
-  
-}
-
-/* 
- * Initialize the hashState_sd.
- */
-int init_sd(hashState_sd *state, int hashbitlen) {
-  int r;
-  char *init;
-
-#ifndef NO_PRECOMPUTED_IV
-//  if (hashbitlen == 224)
-//    r=InitIV(state, hashbitlen, IV_224);
-//  else if (hashbitlen == 256)
-//    r=InitIV(state, hashbitlen, IV_256);
-//  else if (hashbitlen == 384)
-//    r=InitIV(state, hashbitlen, IV_384);
-//  else
-  if (hashbitlen == 512)
-    r = InitIV(state, hashbitlen, IV_512);
-  else
-#endif
-    {
-      /* 
-       * Nonstandart length: IV is not precomputed.
-       */
-      r=InitIV(state, hashbitlen, NULL);
-      if (r != 0)
-        return r;
-      
-      init = malloc(state->blocksize);
-      memset(init, 0, state->blocksize);
-#if defined __STDC__ && __STDC_VERSION__ >= 199901L
-      snprintf(init, state->blocksize, "SIMD-%i v1.1", hashbitlen);
-#else
-      sprintf(init, "SIMD-%i v1.1", hashbitlen);
-#endif
-      SIMD_Compress(state, (unsigned char*) init, 0);
-      free(init);
-    }
-  return r;
-}
-
-int update_sd( hashState_sd *state, const BitSequence *data,
-                      DataLength databitlen )
-{
-  unsigned current;
-  unsigned int bs = state->blocksize;
-  static int align = -1;
-
-  if (align == -1)
-    align = RequiredAlignment();
-
-#ifdef HAS_64
-  current = state->count & (bs - 1);
-#else
-  current = state->count_low & (bs - 1);
-#endif
-  
-  if ( current & 7 )
-  {
-    // The number of hashed bits is not a multiple of 8.
-    // Very painfull to implement and not required by the NIST API.
-    return 1;
-  }
-
-  while ( databitlen > 0 )
-  {
-    if ( IS_ALIGNED(data,align) && current == 0 && databitlen >= bs )
-    {
-       // We can hash the data directly from the input buffer.
-      SIMD_Compress(state, data, 0);
-      databitlen -= bs;
-      data += bs/8;
-      IncreaseCounter(state, bs);
-    }
-    else
-    {
-       // Copy a chunk of data to the buffer
-      unsigned int len = bs - current;
-      if ( databitlen < len )
-      {
-        memcpy( state->buffer+current/8, data, (databitlen+7)/8 );
-        IncreaseCounter( state, databitlen );        
-        return 0;
-      }
-      else
-      {
-        memcpy( state->buffer+current/8, data, len/8 );
-        IncreaseCounter( state,len );
-        databitlen -= len;
-        data += len/8;
-        current = 0;
-        SIMD_Compress( state, state->buffer, 0 );
-      }
-    }
-  }
-  return 0;
-}
-
-int final_sd( hashState_sd *state, BitSequence *hashval )
-{
-#ifdef HAS_64
-  uint64_t l;
-  int current = state->count & (state->blocksize - 1);
-#else
-  uint32_t l;
-  int current = state->count_low & (state->blocksize - 1);
-#endif
-  unsigned int i;
-  BitSequence bs[64];
-  int isshort = 1;
-
-  // If there is still some data in the buffer, hash it
-  if ( current )
-  {
-    // We first need to zero out the end of the buffer.
-    if ( current & 7 )
-    {
-      BitSequence mask = 0xff >> ( current & 7 );
-      state->buffer[current/8] &= ~mask;
-    }
-    current = ( current+7 ) / 8;
-    memset( state->buffer+current, 0, state->blocksize/8 - current );
-    SIMD_Compress( state, state->buffer, 0 );
-  }
-
-  //* Input the message length as the last block
-  memset( state->buffer, 0, state->blocksize / 8 );
-#ifdef HAS_64
-  l = state->count;
-  for ( i=0; i<8; i++ )
-  {
-    state->buffer[i] = l & 0xff;
-    l >>= 8;
-  }
-  if ( state->count < 16384 )
-    isshort = 2;
-#else
-  l = state->count_low;
-  for ( i=0; i<4; i++ )
-  {
-    state->buffer[i] = l & 0xff;
-    l >>= 8;
-  }
-  l = state->count_high;
-  for ( i=0; i<4; i++ )
-  {
-    state->buffer[4+i] = l & 0xff;
-    l >>= 8;
-  }
-  if ( state->count_high == 0 && state->count_low < 16384 )
-    isshort = 2;
-#endif
-
-  SIMD_Compress( state, state->buffer, isshort );
-  
-  // Decode the 32-bit words into a BitSequence
-  for ( i=0; i < 2*state->n_feistels; i++ )
-  {
-    u32 x = state->A[i];
-    bs[4*i  ] = x&0xff;
-    x >>= 8;
-    bs[4*i+1] = x&0xff;
-    x >>= 8;
-    bs[4*i+2] = x&0xff;
-    x >>= 8;
-    bs[4*i+3] = x&0xff;
-  }
-
-  memcpy( hashval, bs, state->hashbitlen / 8 );
-  if ( state->hashbitlen % 8 )
-  {
-    BitSequence mask = 0xff << ( 8 - (state->hashbitlen % 8) );
-    hashval[state->hashbitlen/8 + 1] = bs[state->hashbitlen/8 + 1] & mask;
-  }
-  return 0;
-}
-
-int update_final_sd( hashState_sd *state, BitSequence *hashval,
-                            const BitSequence *data, DataLength databitlen )
-{
-  int current, i;
-  unsigned int bs = state->blocksize;
-  static int align = -1;
-  BitSequence out[64];
-  int isshort = 1;
-  uint64_t l;
-
-  if (align == -1)
-    align = RequiredAlignment();
-
-#ifdef HAS_64
-  current = state->count & (bs - 1);
-#else
-  current = state->count_low & (bs - 1);
-#endif
-
-  if ( current & 7 )
-  {
-    // The number of hashed bits is not a multiple of 8.
-    // Very painfull to implement and not required by the NIST API.
-    return 1;
-  }
-
-  while ( databitlen > 0 )
-  {
-    if ( IS_ALIGNED(data,align) && current == 0 && databitlen >= bs )
-    {
-       // We can hash the data directly from the input buffer.
-      SIMD_Compress(state, data, 0);
-      databitlen -= bs;
-      data += bs/8;
-      IncreaseCounter(state, bs);
-    }
-    else
-    {
-       // Copy a chunk of data to the buffer
-      unsigned int len = bs - current;
-      if ( databitlen < len )
-      {
-        memcpy( state->buffer+current/8, data, (databitlen+7)/8 );
-        IncreaseCounter( state, databitlen );
-        break;
-      }
-      else
-      {
-        memcpy( state->buffer+current/8, data, len/8 );
-        IncreaseCounter( state,len );
-        databitlen -= len;
-        data += len/8;
-        current = 0;
-        SIMD_Compress( state, state->buffer, 0 );
-      }
-    }
-  }
-
-  current = state->count & (state->blocksize - 1);
-
-  // If there is still some data in the buffer, hash it
-  if ( current )
-  {
-    // We first need to zero out the end of the buffer.
-    if ( current & 7 )
-    {
-      BitSequence mask = 0xff >> ( current & 7 );
-      state->buffer[current/8] &= ~mask;
-    }
-    current = ( current+7 ) / 8;
-    memset( state->buffer+current, 0, state->blocksize/8 - current );
-    SIMD_Compress( state, state->buffer, 0 );
-  }
-
-  //* Input the message length as the last block
-  memset( state->buffer, 0, state->blocksize / 8 );
-  l = state->count;
-  for ( i=0; i<8; i++ )
-  {
-    state->buffer[i] = l & 0xff;
-    l >>= 8;
-  }
-  if ( state->count < 16384 )
-    isshort = 2;
-
-  SIMD_Compress( state, state->buffer, isshort );
-
-  // Decode the 32-bit words into a BitSequence
-  for ( i=0; i < 2*state->n_feistels; i++ )
-  {
-    u32 x = state->A[i];
-    out[4*i  ] = x & 0xff;
-    x >>= 8;
-    out[4*i+1] = x & 0xff;
-    x >>= 8;
-    out[4*i+2] = x & 0xff;
-    x >>= 8;
-    out[4*i+3] = x & 0xff;
-  }
-
-  memcpy( hashval, out, state->hashbitlen / 8 );
-  if ( state->hashbitlen % 8 )
-  {
-    BitSequence mask = 0xff << ( 8 - (state->hashbitlen % 8) );
-    hashval[state->hashbitlen/8 + 1] = out[state->hashbitlen/8 + 1] & mask;
-  }
-  return 0;
-}
-
-int simd_full( hashState_sd *state, BitSequence *hashval,
-                            const BitSequence *data, DataLength databitlen )
-{
- 
-
-  InitIV( state, 512, IV_512 );
- 
-  int current, i;
-  unsigned int bs = state->blocksize;
-  static int align = -1;
-  BitSequence out[64];
-  int isshort = 1;
-  uint64_t l;
-
-  if (align == -1)
-    align = RequiredAlignment();
-
-#ifdef HAS_64
-  current = state->count & (bs - 1);
-#else
-  current = state->count_low & (bs - 1);
-#endif
-
-  if ( current & 7 )
-  {
-    // The number of hashed bits is not a multiple of 8.
-    // Very painfull to implement and not required by the NIST API.
-    return 1;
-  }
-
-  while ( databitlen > 0 )
-  {
-    if ( IS_ALIGNED(data,align) && current == 0 && databitlen >= bs )
-    {
-       // We can hash the data directly from the input buffer.
-      SIMD_Compress(state, data, 0);
-      databitlen -= bs;
-      data += bs/8;
-      IncreaseCounter(state, bs);
-    }
-    else
-    {
-       // Copy a chunk of data to the buffer
-      unsigned int len = bs - current;
-      if ( databitlen < len )
-      {
-        memcpy( state->buffer+current/8, data, (databitlen+7)/8 );
-        IncreaseCounter( state, databitlen );
-        break;
-      }
-      else
-      {
-        memcpy( state->buffer+current/8, data, len/8 );
-        IncreaseCounter( state,len );
-        databitlen -= len;
-        data += len/8;
-        current = 0;
-        SIMD_Compress( state, state->buffer, 0 );
-      }
-    }
-  }
-
-  current = state->count & (state->blocksize - 1);
-
-  // If there is still some data in the buffer, hash it
-  if ( current )
-  {
-    // We first need to zero out the end of the buffer.
-    if ( current & 7 )
-    {
-      BitSequence mask = 0xff >> ( current & 7 );
-      state->buffer[current/8] &= ~mask;
-    }
-    current = ( current+7 ) / 8;
-    memset( state->buffer+current, 0, state->blocksize/8 - current );
-    SIMD_Compress( state, state->buffer, 0 );
-  }
-
-  //* Input the message length as the last block
-  memset( state->buffer, 0, state->blocksize / 8 );
-  l = state->count;
-  for ( i=0; i<8; i++ )
-  {
-    state->buffer[i] = l & 0xff;
-    l >>= 8;
-  }
-  if ( state->count < 16384 )
-    isshort = 2;
-
-  SIMD_Compress( state, state->buffer, isshort );
-
-  // Decode the 32-bit words into a BitSequence
-  for ( i=0; i < 2*state->n_feistels; i++ )
-  {
-    u32 x = state->A[i];
-    out[4*i  ] = x & 0xff;
-    x >>= 8;
-    out[4*i+1] = x & 0xff;
-    x >>= 8;
-    out[4*i+2] = x & 0xff;
-    x >>= 8;
-    out[4*i+3] = x & 0xff;
-  }
-
-  memcpy( hashval, out, state->hashbitlen / 8 );
-  if ( state->hashbitlen % 8 )
-  {
-    BitSequence mask = 0xff << ( 8 - (state->hashbitlen % 8) );
-    hashval[state->hashbitlen/8 + 1] = out[state->hashbitlen/8 + 1] & mask;
-  }
-  return 0;
-}
-
-#endif
-

algo/simd/nist.h

@@ -1,64 +0,0 @@
-#ifndef __NIST_H__
-#define __NIST_H__
-
-/*define data alignment for different C compilers*/
-#if defined(__GNUC__)
-#define DATA_ALIGN(x) x __attribute__((aligned(16)))
-#else
-#define DATA_ALIGN(x) __declspec(align(16)) x
-#endif
-
-#include "simd-compat.h"
-#include "compat/sha3-defs.h"
-/*
- * NIST API Specific types.
- */
-
-typedef struct {
-  unsigned int hashbitlen;
-  unsigned int blocksize;
-  unsigned int n_feistels;
-
-#ifdef HAS_64
-  uint64_t count;
-#else
-  uint32_t count_low;
-  uint32_t count_high;
-#endif
-
-  DATA_ALIGN(uint32_t A[32]);
-  uint32_t *B;
-  uint32_t *C;
-  uint32_t *D;
-  DATA_ALIGN(unsigned char buffer[128]);
-  
-} hashState_sd;
-
-/* 
- * NIST API
- */
-
-int init_sd(hashState_sd *state, int hashbitlen);
-
-int update_sd(hashState_sd *state, const BitSequence *data, DataLength databitlen);
-
-int final_sd(hashState_sd *state, BitSequence *hashval);
-
-int update_final_sd( hashState_sd *state, BitSequence *hashval,
-                            const BitSequence *data, DataLength databitlen );
-
-int simd_full( hashState_sd *state, BitSequence *hashval,
-               const BitSequence *data, DataLength databitlen );
-
-/* 
- * Internal API
- */
-
-//int SupportedLength(int hashbitlen);
-int RequiredAlignment(void);
-void SIMD_Compress(hashState_sd * state, const unsigned char *M, int final);
-
-void fft128_natural(fft_t *a, unsigned char *x);
-void fft256_natural(fft_t *a, unsigned char *x);
-
-#endif

algo/simd/simd-compat.h

@@ -1,198 +0,0 @@
-#ifndef __SIMD_COMPAT_H__
-#define __SIMD_COMPAT_H__
-
-#include <limits.h>
-
-
-/* 
- * This file desfines some helper function for cross-platform compatibility.
- */
-
-#if defined __GNUC_PREREQ && (! defined __STRICT_ANSI__)
-#define GNU_EXT
-#endif
-
-/*
- * First define some integer types.
- */
-
-#if defined __STDC__ && __STDC_VERSION__ >= 199901L
-
-/*
- * On C99 implementations, we can use <stdint.h> to get an exact 32-bit
- * type, if any, or otherwise use a wider type.
- */
-
-#include <stdint.h>
-#include "compat/brg_types.h"
-
-#define C32(x)    ((u32)(x))
-
-#define HAS_64  1
-
-#else
-
-/*
- * On non-C99 systems, we use "unsigned int" if it is wide enough,
- * "unsigned long" otherwise. This supports all "reasonable" architectures.
- * We have to be cautious: pre-C99 preprocessors handle constants
- * differently in '#if' expressions. Hence the shifts to test UINT_MAX.
- */
-
-#if ((UINT_MAX >> 11) >> 11) >= 0x3FF
-
-typedef unsigned int u32;
-
-#define C32(x)    ((u32)(x ## U))
-
-#else
-
-typedef unsigned long u32;
-
-#define C32(x)    ((u32)(x ## UL))
-
-#endif
-
-/*
- * We want a 64-bit type. We use "unsigned long" if it is wide enough (as
- * is common on 64-bit architectures such as AMD64, Alpha or Sparcv9),
- * "unsigned long long" otherwise, if available. We use ULLONG_MAX to
- * test whether "unsigned long long" is available; we also know that
- * gcc features this type, even if the libc header do not know it.
- */
-
-#if ((ULONG_MAX >> 31) >> 31) >= 3
-
-typedef unsigned long u64;
-
-#define HAS_64  1
-
-#elif ((ULLONG_MAX >> 31) >> 31) >= 3 || defined __GNUC__
-
-typedef unsigned long long u64;
-
-#define HAS_64  1
-
-#else
-
-/*
- * No 64-bit type...
- */
-
-#endif
-
-#endif
-
-
-/*
- * fft_t should be at least 16 bits wide.
- * using short int will require less memory, but int is faster...
- */
-
-typedef int fft_t;
-
-
-/*
- * Implementation note: some processors have specific opcodes to perform
- * a rotation. Recent versions of gcc recognize the expression above and
- * use the relevant opcodes, when appropriate.
- */
-
-#define T32(x)    ((x) & C32(0xFFFFFFFF))
-#define ROTL32(x, n)   T32(((x) << (n)) | ((x) >> (32 - (n))))
-#define ROTR32(x, n)   ROTL32(x, (32 - (n)))
-
-
-
-/*
- * The macro MAYBE_INLINE expands to an inline qualifier, is available.
- */
-
-#if (defined __STDC__ && __STDC_VERSION__ >= 199901L) || defined GNU_EXT
-#define MAYBE_INLINE static inline
-#elif defined _MSC_VER
-#define MAYBE_INLINE __inline
-#else
-#define MAYBE_INLINE
-#endif
-
-
-/*  */
-
-#if defined __GNUC__ && ( defined __i386__ || defined __x86_64__ )
-
-#define rdtsc()                                                         \
-  ({                                                                    \
-    u32 lo, hi;                                                         \
-    __asm__ __volatile__ (      /* serialize */                         \
-                          "xorl %%eax,%%eax \n        cpuid"            \
-                          ::: "%rax", "%rbx", "%rcx", "%rdx");          \
-    /* We cannot use "=A", since this would use %rax on x86_64 */       \
-    __asm__ __volatile__ ("rdtsc" : "=a" (lo), "=d" (hi));              \
-    (u64)hi << 32 | lo;                                                 \
-  })                                                                    \
-
-#elif defined _MSC_VER && (defined _M_IX86 || defined _M_X64)
-
-#define rdtsc __rdtsc
-
-#endif
-
-/* 
- * The IS_ALIGNED macro tests if a char* pointer is aligned to an
- * n-bit boundary.
- * It is defined as false on unknown architectures.
- */
-
-
-#define CHECK_ALIGNED(p,n) ((((unsigned char *) (p) - (unsigned char *) NULL) & ((n)-1)) == 0)
-
-#if defined __i386__ || defined __x86_64 || defined _M_IX86 || defined _M_X64
-/*
- * Unaligned 32-bit access are not expensive on x86 so we don't care
- */
-#define IS_ALIGNED(p,n)    (n<=4 || CHECK_ALIGNED(p,n))
-
-#elif defined __sparcv9 || defined __sparc || defined __arm || \
-      defined __ia64 || defined __ia64__ || \
-      defined __itanium__ || defined __M_IA64 || \
-      defined __powerpc__ || defined __powerpc
-#define IS_ALIGNED(p,n)    CHECK_ALIGNED(p,n)
-
-#else
-/* 
- * Unkonwn architecture: play safe
- */
-#define IS_ALIGNED(p,n)    0
-#endif
-
-
-
-/* checks for endianness */
-
-#if defined (__linux__) || defined (__GLIBC__)
-#  include <endian.h>
-#elif defined (__FreeBSD__)
-#  include <machine/endian.h> 
-#elif defined (__OpenBSD__)
-#  include <sys/endian.h>
-#endif
-
-#ifdef __BYTE_ORDER
-
-#  if __BYTE_ORDER == __LITTLE_ENDIAN
-#    define SIMD_LITTLE_ENDIAN
-#  elif __BYTE_ORDER == __BIG_ENDIAN
-#    define SIMD_BIG_ENDIAN
-#  endif
-
-#else
-
-#  if defined __i386__ || defined __x86_64 || defined _M_IX86 || defined _M_X64
-#    define SIMD_LITTLE_ENDIAN
-#  endif
-
-#endif
-
-
-#endif

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

@@ -1,7 +1,6 @@
 #ifndef SIMD_HASH_2WAY_H__
 #define SIMD_HASH_2WAY_H__ 1
 
-#include "simd-compat.h"
 #include "simd-utils.h"
 
 #if defined(__SSE2__) || defined (__ARM_NEON)
@@ -34,7 +33,7 @@ typedef struct
   unsigned int hashbitlen;
   unsigned int blocksize;
   unsigned int n_feistels;
-} simd512_2way_context __attribute__((aligned(128)));
+} simd512_2way_context __attribute__((aligned(64)));
 #define simd_2way_context simd512_2way_context
 
 // databitlen is bits

algo/simd/vector.c

@@ -1,948 +0,0 @@
-#include <stdlib.h>
-#include <stdio.h>
-
-#include "nist.h"
-#include "vector.h"
-
-
-//#if defined(__SSE2__) || defined(__ARM_NEON)
-#if defined(__SSE2__)
-
-#define PRINT_SOME 0
-
-/*
-int SupportedLength(int hashbitlen) {
-  if (hashbitlen <= 0 || hashbitlen > 512)
-    return 0;
-  else
-    return 1;
-}
-*/
-
-int RequiredAlignment(void) {
-  return 16;
-}
-
-static const union cv V128 = CV(128);
-static const union cv V255 = CV(255);
-static const union cv V257 = CV(257);
-static const union cv8  V0 = CV(0);
-
-
-/*
- * Reduce modulo 257; result is in [-127; 383]
- * REDUCE(x) := (x&255) - (x>>8)
- */
-#define REDUCE(x)                               \
-  v16_sub(v16_and(x, V255.v16), v16_shift_r (x, 8))
-
-/*
- * Reduce from [-127; 383] to [-128; 128]
- * EXTRA_REDUCE_S(x) := x<=128 ? x : x-257
- */
-#define EXTRA_REDUCE_S(x)                       \
-  v16_sub(x, v16_and(V257.v16, v16_cmp(x, V128.v16)))
- 
-/*
- * Reduce modulo 257; result is in [-128; 128]
- */
-#define REDUCE_FULL_S(x)                        \
-  EXTRA_REDUCE_S(REDUCE(x))
-
-#define DO_REDUCE(i)                            \
-  X(i) = REDUCE(X(i))
-
-#define DO_REDUCE_FULL_S(i)                     \
-  do {                                          \
-    X(i) = REDUCE(X(i));                        \
-    X(i) = EXTRA_REDUCE_S(X(i));                \
-  } while(0)
-
-#define MAYBE_VOLATILE
-
-MAYBE_INLINE void fft64(void *a) {
-
-  v16* const A = a;
-
-  register v16 X0, X1, X2, X3, X4, X5, X6, X7;
-/*
-#if V16_SIZE == 8
-#define X(i) A[i]
-#elif V16_SIZE == 4
-#define X(i) A[2*i]
-#endif
-*/
-#define X(i) X##i
-
-  X0 = A[0];
-  X1 = A[1];
-  X2 = A[2];
-  X3 = A[3];
-  X4 = A[4];
-  X5 = A[5];
-  X6 = A[6];
-  X7 = A[7];
-
-#define DO_REDUCE(i)                            \
-  X(i) = REDUCE(X(i))
-
-  /*
-   * Begin with 8 parallels DIF FFT_8
-   *
-   * FFT_8 using w=4 as 8th root of unity
-   *  Unrolled decimation in frequency (DIF) radix-2 NTT.
-   *  Output data is in revbin_permuted order.
-   */
-
-  static const int w[] = {0, 2, 4, 6};
-  //  v16 *Twiddle = (v16*)FFT64_Twiddle;
-
-#define BUTTERFLY(i,j,n)                                \
-  do {                                                  \
-    MAYBE_VOLATILE v16 v = X(j);                              \
-    X(j) =  v16_add(X(i), X(j));                        \
-    if (n)                                              \
-      X(i) = v16_shift_l(v16_sub(X(i), v), w[n]);       \
-    else                                                \
-      X(i) = v16_sub(X(i), v);                          \
-  } while(0)
-
-  BUTTERFLY(0, 4, 0);
-  BUTTERFLY(1, 5, 1);
-  BUTTERFLY(2, 6, 2);
-  BUTTERFLY(3, 7, 3);
-  
-  DO_REDUCE(2);
-  DO_REDUCE(3);
-  
-  BUTTERFLY(0, 2, 0);
-  BUTTERFLY(4, 6, 0);
-  BUTTERFLY(1, 3, 2);
-  BUTTERFLY(5, 7, 2);
-  
-  DO_REDUCE(1);
-  
-  BUTTERFLY(0, 1, 0);
-  BUTTERFLY(2, 3, 0);
-  BUTTERFLY(4, 5, 0);
-  BUTTERFLY(6, 7, 0);
-  
-  /* We don't need to reduce X(7) */
-  DO_REDUCE_FULL_S(0);
-  DO_REDUCE_FULL_S(1);
-  DO_REDUCE_FULL_S(2);
-  DO_REDUCE_FULL_S(3);
-  DO_REDUCE_FULL_S(4);
-  DO_REDUCE_FULL_S(5);
-  DO_REDUCE_FULL_S(6);
-    
-#undef BUTTERFLY
-
-  /*
-   * Multiply by twiddle factors
-   */
-
-  X(6) = v16_mul(X(6), FFT64_Twiddle[0].v16);
-  X(5) = v16_mul(X(5), FFT64_Twiddle[1].v16);
-  X(4) = v16_mul(X(4), FFT64_Twiddle[2].v16);
-  X(3) = v16_mul(X(3), FFT64_Twiddle[3].v16);
-  X(2) = v16_mul(X(2), FFT64_Twiddle[4].v16);
-  X(1) = v16_mul(X(1), FFT64_Twiddle[5].v16);
-  X(0) = v16_mul(X(0), FFT64_Twiddle[6].v16);
-
-  /*
-   * Transpose the FFT state with a revbin order permutation
-   * on the rows and the column.
-   * This will make the full FFT_64 in order.
-   */
-
-#define INTERLEAVE(i,j)                          \
-  do {                                           \
-    v16 t1= X(i);                                \
-    v16 t2= X(j);                                \
-    X(i) = v16_interleavel(t1, t2);              \
-    X(j) = v16_interleaveh(t1, t2);              \
-  } while(0)
-
-  INTERLEAVE(1, 0);
-  INTERLEAVE(3, 2);
-  INTERLEAVE(5, 4);
-  INTERLEAVE(7, 6);
-
-  INTERLEAVE(2, 0);
-  INTERLEAVE(3, 1);
-  INTERLEAVE(6, 4);
-  INTERLEAVE(7, 5);
-
-  INTERLEAVE(4, 0);
-  INTERLEAVE(5, 1);
-  INTERLEAVE(6, 2);
-  INTERLEAVE(7, 3);
-
-#undef INTERLEAVE
-
-  /*
-   * Finish with 8 parallels DIT FFT_8
-   *
-   * FFT_8 using w=4 as 8th root of unity
-   *  Unrolled decimation in time (DIT) radix-2 NTT.
-   *  Intput data is in revbin_permuted order.
-   */
-  
-#define BUTTERFLY(i,j,n)                                \
-  do {                                                  \
-    MAYBE_VOLATILE v16 u = X(j);                              \
-    if (n)                                              \
-      X(i) = v16_shift_l(X(i), w[n]);                   \
-    X(j) = v16_sub(X(j), X(i));                         \
-    X(i) = v16_add(u, X(i));                            \
-  } while(0)
-
-  DO_REDUCE(0);
-  DO_REDUCE(1);
-  DO_REDUCE(2);
-  DO_REDUCE(3);
-  DO_REDUCE(4);
-  DO_REDUCE(5);
-  DO_REDUCE(6);
-  DO_REDUCE(7);
-  
-  BUTTERFLY(0, 1, 0);
-  BUTTERFLY(2, 3, 0);
-  BUTTERFLY(4, 5, 0);
-  BUTTERFLY(6, 7, 0);
-  
-  BUTTERFLY(0, 2, 0);
-  BUTTERFLY(4, 6, 0);
-  BUTTERFLY(1, 3, 2);
-  BUTTERFLY(5, 7, 2);
-  
-  DO_REDUCE(3);
-  
-  BUTTERFLY(0, 4, 0);
-  BUTTERFLY(1, 5, 1);
-  BUTTERFLY(2, 6, 2);
-  BUTTERFLY(3, 7, 3);
-  
-  DO_REDUCE_FULL_S(0);
-  DO_REDUCE_FULL_S(1);
-  DO_REDUCE_FULL_S(2);
-  DO_REDUCE_FULL_S(3);
-  DO_REDUCE_FULL_S(4);
-  DO_REDUCE_FULL_S(5);
-  DO_REDUCE_FULL_S(6);
-  DO_REDUCE_FULL_S(7);
-  
-#undef BUTTERFLY
-
-  A[0] = X0;
-  A[1] = X1;
-  A[2] = X2;
-  A[3] = X3;
-  A[4] = X4;
-  A[5] = X5;
-  A[6] = X6;
-  A[7] = X7;
-
-#undef X
-
-}
-
-
-MAYBE_INLINE void fft128(void *a) {
-
-  int i;
-
-  // Temp space to help for interleaving in the end
-  v16 B[8];
-
-  v16 *A = (v16*) a;
-  //  v16 *Twiddle = (v16*)FFT128_Twiddle;
-
-  /* Size-2 butterflies */
-
-  for (i = 0; i<8; i++) {
-    B[i]   = v16_add(A[i], A[i+8]);
-    B[i]   = REDUCE_FULL_S(B[i]);
-    A[i+8] = v16_sub(A[i], A[i+8]);
-    A[i+8] = REDUCE_FULL_S(A[i+8]);
-    A[i+8] = v16_mul(A[i+8], FFT128_Twiddle[i].v16);
-    A[i+8] = REDUCE_FULL_S(A[i+8]);
-  }
-
-  fft64(B);
-  fft64(A+8);
-
-  /* Transpose (i.e. interleave) */
-
-  for (i=0; i<8; i++) {
-    A[2*i]   = v16_interleavel (B[i], A[i+8]);
-    A[2*i+1] = v16_interleaveh (B[i], A[i+8]);
-  }
-}
-
-#ifdef v16_broadcast
-/* Compute the FFT using a table
- * The function works if the value of the message is smaller 
- * than 2^14.
- */
-void fft128_msg_final(short *a, const unsigned char *x) {
-
-  static const union cv FFT128_Final_Table[] = {
-    {{   1, -211,   60,  -67,    2,   92, -137,  123}},
-    {{   2,  118,   45,  111,   97,  -46,   49, -106}},
-    {{   4,  -73,  -17,  -11,    8,  111,  -34,  -22}},
-    {{ -68,   -4,   76,  -25,   96,  -96,  -68,   -9}},
-    {{  16,  -35,  -68,  -44,   32,  -70, -136,  -88}},
-    {{   0, -124,   17,   12,   -6,   57,   47,   -8}},
-    {{  64,  117,  -15,   81,  128,  -23,  -30,  -95}},
-    {{ -68,  -53,  -52,  -70,  -10, -117,   77,   21}},
-    {{  -1,  -46,  -60,   67,   -2,  -92, -120, -123}},
-    {{  -2, -118,  -45, -111,  -97,   46,  -49,  106}},
-    {{  -4,   73,   17,   11,   -8, -111,   34,   22}},
-    {{  68,    4,  -76,   25,  -96,   96,   68,    9}},
-    {{ -16, -222,   68,   44,  -32,   70, -121,   88}},
-    {{   0,  124,  -17,  -12,    6,  -57,  -47,    8}},
-    {{ -64, -117,   15,  -81, -128, -234,   30,   95}},
-    {{  68,   53,   52,   70,   10,  117,  -77,  -21}},
-    {{-118,  -31,  116,  -61,   21,  -62,  -25, -122}},
-    {{-101,  107,  -45,  -95,   -8,    3,  101,  -34}},
-    {{  42, -124,  -50,   13,   84,    9, -100, -231}},
-    {{ -79,  -53,   82,   65,  -81,   47,   61,  107}},
-    {{ -89, -239,   57, -205, -178,   36, -143,  104}},
-    {{-126,  113,   33,  111,  103, -109,   65, -114}},
-    {{ -99,   72,  -29,  -49, -198, -113,  -58,  -98}},
-    {{   8,  -27, -106,  -30,  111,    6,   10, -108}},
-    {{-139,   31, -116, -196,  -21,   62,   25, -135}},
-    {{ 101, -107,   45,   95,    8,   -3, -101,   34}},
-    {{ -42, -133,   50,  -13,  -84,   -9,  100,  -26}},
-    {{  79,   53,  -82,  -65,   81,  -47,  -61, -107}},
-    {{-168,  -18,  -57,  -52,  -79,  -36, -114, -104}},
-    {{ 126, -113,  -33, -111, -103,  109,  -65,  114}},
-    {{  99,  -72, -228,   49,  -59,  113,   58, -159}},
-    {{  -8,   27,  106,   30, -111,   -6,  -10,  108}}
-  };
-
-  //  v16 *Table = (v16*)FFT128_Final_Table;
-  v16 *A = (v16*) a;
-  v16 msg1 = v16_broadcast(x[0]>128?x[0]-257:x[0]);
-  v16 msg2 = v16_broadcast(x[1]>128?x[1]-257:x[1]);
-  // v16 msg2 = v16_broadcast(x[1]);
-
-#if 0
-  int i;
-  for (i=0; i<16; i++) {
-    v16 tmp = v16_mul(FFT128_Final_Table[2*i].v16  , msg2);
-    v16 sum = v16_add(FFT128_Final_Table[2*i+1].v16, msg1);
-    sum = v16_add(sum, tmp);
-    A[i] = REDUCE_FULL_S(sum);
-  }
-
-#else
-
-#define FFT_FINAL(i)                                           \
-  v16 tmp##i = v16_mul(FFT128_Final_Table[2*i].v16, msg2);     \
-  v16 sum##i = v16_add(FFT128_Final_Table[2*i+1].v16, msg1);   \
-  sum##i = v16_add(sum##i, tmp##i);                            \
-  A[i] = REDUCE_FULL_S(sum##i);
-
-  FFT_FINAL(0)
-  FFT_FINAL(1)
-  FFT_FINAL(2)
-  FFT_FINAL(3)
-  FFT_FINAL(4)
-  FFT_FINAL(5)
-  FFT_FINAL(6)
-  FFT_FINAL(7)
-  FFT_FINAL(8)
-  FFT_FINAL(9)
-  FFT_FINAL(10)
-  FFT_FINAL(11)
-  FFT_FINAL(12)
-  FFT_FINAL(13)
-  FFT_FINAL(14)
-  FFT_FINAL(15)
-
-#endif
-
-}
-#endif
-
-void fft128_msg(short *a, const unsigned char *x, int final) {
-
-  static const union cv Tweak =
-    {{0,0,0,0,0,0,0,1}};
-  static const union cv FinalTweak =
-    {{0,0,0,0,0,1,0,1}};
-
-
-  v8  *X = (v8*)  x;
-  v16 *A = (v16*) a;
-  //  v16 *Twiddle = (v16*)FFT128_Twiddle;
-
-#define UNPACK(i)                                      \
-  do {                                                 \
-    v8 t = X[i];                                       \
-    A[2*i]   = v8_mergel(t, V0.v8);                    \
-    A[2*i+8] = v16_mul(A[2*i], FFT128_Twiddle[2*i].v16);          \
-    A[2*i+8] = REDUCE(A[2*i+8]);                       \
-    A[2*i+1] = v8_mergeh(t, V0.v8);                    \
-    A[2*i+9] = v16_mul(A[2*i+1], FFT128_Twiddle[2*i+1].v16);      \
-    A[2*i+9] = REDUCE(A[2*i+9]);                       \
-  } while(0)
-
-
-  /* 
-   * This allows to tweak the last butterflies to introduce X^127
-   */
-#define UNPACK_TWEAK(i,tw)                             \
-  do {                                                 \
-    v8 t = X[i];                                       \
-    v16 tmp;                                           \
-    A[2*i]   = v8_mergel(t, V0.v8);                    \
-    A[2*i+8] = v16_mul(A[2*i], FFT128_Twiddle[2*i].v16);          \
-    A[2*i+8] = REDUCE(A[2*i+8]);                       \
-    tmp      = v8_mergeh(t, V0.v8);                    \
-    A[2*i+1] = v16_add(tmp, tw);                               \
-    A[2*i+9] = v16_mul(v16_sub(tmp, tw), FFT128_Twiddle[2*i+1].v16);      \
-    A[2*i+9] = REDUCE(A[2*i+9]);                       \
-  } while(0)
-
-  UNPACK(0);
-  UNPACK(1);
-  UNPACK(2);
-  if (final)
-    UNPACK_TWEAK(3, FinalTweak.v16);
-  else
-    UNPACK_TWEAK(3, Tweak.v16);
-
-#undef UNPACK
-#undef UNPACK_TWEAK
-
-  fft64(a);
-  fft64(a+64);
-}
-
-#if 0
-void fft128_msg(short *a, const unsigned char *x, int final) {
-
-  for (int i=0; i<64; i++)
-    a[i] = x[i];
-
-  for (int i=64; i<128; i++)
-    a[i] = 0;
-
-  a[127] = 1;
-  a[125] = final? 1: 0;
-
-  fft128(a);
-}
-#endif
-
-void fft256_msg(short *a, const unsigned char *x, int final) {
-
-  static const union cv Tweak =
-    {{0,0,0,0,0,0,0,1}};
-  static const union cv FinalTweak =
-    {{0,0,0,0,0,1,0,1}};
-
-
-  v8  *X = (v8*)  x;
-  v16 *A = (v16*) a;
-  //  v16 *Twiddle = (v16*)FFT256_Twiddle;
-
-#define UNPACK(i)                                       \
-  do {                                                  \
-    v8 t      = X[i];                                   \
-    A[2*i]    = v8_mergel(t, V0.v8);                    \
-    A[2*i+16] = v16_mul(A[2*i], FFT256_Twiddle[2*i].v16);          \
-    A[2*i+16] = REDUCE(A[2*i+16]);                      \
-    A[2*i+1]  = v8_mergeh(t, V0.v8);                    \
-    A[2*i+17] = v16_mul(A[2*i+1], FFT256_Twiddle[2*i+1].v16);      \
-    A[2*i+17] = REDUCE(A[2*i+17]);                       \
-  } while(0)
-
-
-  /* 
-   * This allows to tweak the last butterflies to introduce X^127
-   */
-#define UNPACK_TWEAK(i,tw)                              \
-  do {                                                  \
-    v8 t = X[i];                                        \
-    v16 tmp;                                            \
-    A[2*i]    = v8_mergel(t, V0.v8);                    \
-    A[2*i+16] = v16_mul(A[2*i], FFT256_Twiddle[2*i].v16);          \
-    A[2*i+16] = REDUCE(A[2*i+16]);                       \
-    tmp       = v8_mergeh(t, V0.v8);                    \
-    A[2*i+1]  = v16_add(tmp, tw);                               \
-    A[2*i+17] = v16_mul(v16_sub(tmp, tw), FFT256_Twiddle[2*i+1].v16);      \
-    A[2*i+17] = REDUCE(A[2*i+17]);                      \
-  } while(0)
-
-  UNPACK(0);
-  UNPACK(1);
-  UNPACK(2);
-  UNPACK(3);
-  UNPACK(4);
-  UNPACK(5);
-  UNPACK(6);
-  if (final)
-    UNPACK_TWEAK(7, FinalTweak.v16);
-  else
-    UNPACK_TWEAK(7, Tweak.v16);
-
-#undef UNPACK
-#undef UNPACK_TWEAK
-
-  fft128(a);
-  fft128(a+128);
-}
-
-
-void rounds(u32* state, const unsigned char* msg, short* fft) {
-  
-  v32* S = (v32*) state;
-  const v32* M = (v32*)msg;
-  volatile v16* W = (v16*)fft;
-
-  register v32 S0, S1, S2, S3;
-  static const union cv code[] = { CV(185), CV(233) };
-
-  S0 = v32_xor(S[0], v32_bswap(M[0]));
-  S1 = v32_xor(S[1], v32_bswap(M[1]));
-  S2 = v32_xor(S[2], v32_bswap(M[2]));
-  S3 = v32_xor(S[3], v32_bswap(M[3]));
-
-#define S(i) S##i
-
-
-/* #define F_0(B, C, D)     ((((C) ^ (D)) & (B)) ^ (D)) */
-/* #define F_1(B, C, D)     (((D) & (C)) | (((D) | (C)) & (B))) */
-
-#define F_0(B, C, D)     v32_xor(v32_and(v32_xor(C,D), B), D)
-#define F_1(B, C, D)     v32_or(v32_and(D, C), v32_and( v32_or(D,C), B))
-
-#define F(a,b,c,fun) F_##fun (a,b,c)
-
-  /*
-   * We split the round function in two halfes
-   * so as to insert some independent computations in between
-   */
-
-#define SUM3_00 1
-#define SUM3_01 2
-#define SUM3_02 3
-#define SUM3_10 2
-#define SUM3_11 3
-#define SUM3_12 1
-#define SUM3_20 3
-#define SUM3_21 1
-#define SUM3_22 2
-
-#define STEP_1(a,b,c,d,w,fun,r,s,z)                             \
-  do {                                                          \
-    if (PRINT_SOME) {                                           \
-      int j;                                                    \
-      v32 ww=w, aa=a, bb=b, cc=c, dd=d;                         \
-      u32 *WW = (void*)&ww;                                     \
-      u32 *AA = (void*)&aa;                                     \
-      u32 *BB = (void*)&bb;                                     \
-      u32 *CC = (void*)&cc;                                     \
-      u32 *DD = (void*)&dd;                                     \
-      for (j=0; j<4; j++) {                                     \
-        printf ("%08x/%2i/%2i[%i]: %08x %08x %08x %08x\n",      \
-                WW[j], r, s, SUM3_##z,                          \
-                AA[j], BB[j], CC[j], DD[j]);                    \
-      }                                                         \
-    }                                                           \
-    TT = F(a,b,c,fun);                                          \
-    a = v32_rotate(a,r);                                        \
-    w = v32_add(w, d);                                          \
-    TT = v32_add(TT, w);                                        \
-    TT = v32_rotate(TT,s);                                      \
-    d = v32_shufxor(a,SUM3_##z);                                \
-  } while(0)
-
-#define STEP_2(a,b,c,d,w,fun,r,s)                               \
-  do {                                                          \
-    d = v32_add(d, TT);                                         \
-  } while(0)
-
-#define STEP(a,b,c,d,w,fun,r,s,z)               \
-  do {                                          \
-    register v32 TT;                            \
-    STEP_1(a,b,c,d,w,fun,r,s,z);                \
-    STEP_2(a,b,c,d,w,fun,r,s);                  \
-  } while(0);
-
-
-#define ROUND(h0,l0,u0,h1,l1,u1,h2,l2,u2,h3,l3,u3,        \
-              fun,r,s,t,u,z,r0)                           \
-  do {                                                    \
-    register v32 W0, W1, W2, W3, TT;                      \
-    W0 = v16_merge##u0(W[h0], W[l0]);                     \
-    W0 = V1632(v16_mul(V3216(W0), code[z].v16));          \
-    STEP_1(S(0), S(1), S(2), S(3), W0, fun, r, s, r0##0); \
-    W1 = v16_merge##u1(W[h1], W[l1]);                     \
-    W1 = V1632(v16_mul(V3216(W1), code[z].v16));          \
-    STEP_2(S(0), S(1), S(2), S(3), W0, fun, r, s);        \
-    STEP_1(S(3), S(0), S(1), S(2), W1, fun, s, t, r0##1); \
-    W2 = v16_merge##u2(W[h2], W[l2]);                     \
-    W2 = V1632(v16_mul(V3216(W2), code[z].v16));          \
-    STEP_2(S(3), S(0), S(1), S(2), W1, fun, s, t);        \
-    STEP_1(S(2), S(3), S(0), S(1), W2, fun, t, u, r0##2); \
-    W3 = v16_merge##u3(W[h3], W[l3]);                     \
-    W3 = V1632(v16_mul(V3216(W3), code[z].v16));          \
-    STEP_2(S(2), S(3), S(0), S(1), W2, fun, t, u);        \
-    STEP_1(S(1), S(2), S(3), S(0), W3, fun, u, r, r0##0); \
-    STEP_2(S(1), S(2), S(3), S(0), W3, fun, u, r);        \
-  } while(0)
-
-
-  /*
-   * 4 rounds with code 185
-   */
-  ROUND(  2, 10, l,  3, 11, l,  0,  8, l,  1,  9, l, 0, 3,  23, 17, 27, 0, 0);
-  ROUND(  3, 11, h,  2, 10, h,  1,  9, h,  0,  8, h, 1, 3,  23, 17, 27, 0, 1);
-  ROUND(  7, 15, h,  5, 13, h,  6, 14, l,  4, 12, l, 0, 28, 19, 22,  7, 0, 2);
-  ROUND(  4, 12, h,  6, 14, h,  5, 13, l,  7, 15, l, 1, 28, 19, 22,  7, 0, 0);
-
-  /*
-   * 4 rounds with code 233
-   */
-  ROUND(  0,  4, h,  1,  5, l,  3,  7, h,  2,  6, l, 0, 29,  9, 15,  5, 1, 1);
-  ROUND(  3,  7, l,  2,  6, h,  0,  4, l,  1,  5, h, 1, 29,  9, 15,  5, 1, 2);
-  ROUND( 11, 15, l,  8, 12, l,  8, 12, h, 11, 15, h, 0,  4, 13, 10, 25, 1, 0);
-  ROUND(  9, 13, h, 10, 14, h, 10, 14, l,  9, 13, l, 1,  4, 13, 10, 25, 1, 1);
-
-
-  /*
-   * 1 round as feed-forward
-   */
-  STEP(S(0), S(1), S(2), S(3), S[0], 0,  4, 13, 20);
-  STEP(S(3), S(0), S(1), S(2), S[1], 0, 13, 10, 21);
-  STEP(S(2), S(3), S(0), S(1), S[2], 0, 10, 25, 22);
-  STEP(S(1), S(2), S(3), S(0), S[3], 0, 25,  4, 20);
-
-  S[0] = S(0);  S[1] = S(1);  S[2] = S(2);  S[3] = S(3);
-
-#undef ROUND
-#undef STEP
-#undef STEP_1
-#undef STEP_2
-}
-
-
-void rounds512(u32* state, const unsigned char* msg, short* fft) {
-  
-  v32* S = (v32*) state;
-  v32* M = (v32*) msg;
-  v16* W = (v16*) fft;
-
-  register v32 S0l, S1l, S2l, S3l;
-  register v32 S0h, S1h, S2h, S3h;
-  static const union cv code[] = { CV(185), CV(233) };
-
-  S0l = v32_xor(S[0], v32_bswap(M[0]));
-  S0h = v32_xor(S[1], v32_bswap(M[1]));
-  S1l = v32_xor(S[2], v32_bswap(M[2]));
-  S1h = v32_xor(S[3], v32_bswap(M[3]));
-  S2l = v32_xor(S[4], v32_bswap(M[4]));
-  S2h = v32_xor(S[5], v32_bswap(M[5]));
-  S3l = v32_xor(S[6], v32_bswap(M[6]));
-  S3h = v32_xor(S[7], v32_bswap(M[7]));
-
-#define S(i) S##i
-
-
-/* #define F_0(B, C, D)     ((((C) ^ (D)) & (B)) ^ (D)) */
-/* #define F_1(B, C, D)     (((D) & (C)) | (((D) | (C)) & (B))) */
-
-#define F_0(B, C, D)     v32_xor(v32_and(v32_xor(C,D), B), D)
-#define F_1(B, C, D)     v32_or(v32_and(D, C), v32_and( v32_or(D,C), B))
-
-#define Fl(a,b,c,fun) F_##fun (a##l,b##l,c##l)
-#define Fh(a,b,c,fun) F_##fun (a##h,b##h,c##h)
-
-  /*
-   * We split the round function in two halfes
-   * so as to insert some independent computations in between
-   */
-
-#define SUM7_00 0
-#define SUM7_01 1
-#define SUM7_02 2
-#define SUM7_03 3
-#define SUM7_04 4
-#define SUM7_05 5
-#define SUM7_06 6
-
-#define SUM7_10 1
-#define SUM7_11 2
-#define SUM7_12 3
-#define SUM7_13 4
-#define SUM7_14 5
-#define SUM7_15 6
-#define SUM7_16 0
-                
-#define SUM7_20 2
-#define SUM7_21 3
-#define SUM7_22 4
-#define SUM7_23 5
-#define SUM7_24 6
-#define SUM7_25 0
-#define SUM7_26 1
-                
-#define SUM7_30 3
-#define SUM7_31 4
-#define SUM7_32 5
-#define SUM7_33 6
-#define SUM7_34 0
-#define SUM7_35 1
-#define SUM7_36 2
-                
-#define SUM7_40 4
-#define SUM7_41 5
-#define SUM7_42 6
-#define SUM7_43 0
-#define SUM7_44 1
-#define SUM7_45 2
-#define SUM7_46 3
-                
-#define SUM7_50 5
-#define SUM7_51 6
-#define SUM7_52 0
-#define SUM7_53 1
-#define SUM7_54 2
-#define SUM7_55 3
-#define SUM7_56 4
-
-#define SUM7_60 6
-#define SUM7_61 0
-#define SUM7_62 1
-#define SUM7_63 2
-#define SUM7_64 3
-#define SUM7_65 4
-#define SUM7_66 5
-
-#define PERM(z,d,a) XCAT(PERM_,XCAT(SUM7_##z,PERM_START))(d,a)
-
-#define PERM_0(d,a) /* XOR 1 */           \
-  do {                                    \
-    d##l = v32_shufxor(a##l,1);           \
-    d##h = v32_shufxor(a##h,1);           \
-  } while(0)
-
-#define PERM_1(d,a) /* XOR 6 */           \
-  do {                                    \
-    d##l = v32_shufxor(a##h,2);           \
-    d##h = v32_shufxor(a##l,2);           \
-  } while(0)
-
-#define PERM_2(d,a) /* XOR 2 */           \
-  do {                                    \
-    d##l = v32_shufxor(a##l,2);           \
-    d##h = v32_shufxor(a##h,2);           \
-  } while(0)
-
-#define PERM_3(d,a) /* XOR 3 */           \
-  do {                                    \
-    d##l = v32_shufxor(a##l,3);           \
-    d##h = v32_shufxor(a##h,3);           \
-  } while(0)
-
-#define PERM_4(d,a) /* XOR 5 */           \
-  do {                                    \
-    d##l = v32_shufxor(a##h,1);           \
-    d##h = v32_shufxor(a##l,1);           \
-  } while(0)
-
-#define PERM_5(d,a) /* XOR 7 */           \
-  do {                                    \
-    d##l = v32_shufxor(a##h,3);           \
-    d##h = v32_shufxor(a##l,3);           \
-  } while(0)
-
-#define PERM_6(d,a) /* XOR 4 */           \
-  do {                                    \
-    d##l = a##h;                          \
-    d##h = a##l;                          \
-  } while(0)
-
-#define STEP_1_(a,b,c,d,w,fun,r,s,z)                            \
-  do {                                                          \
-    if (PRINT_SOME) {                                           \
-      int j;                                                    \
-      v32 ww=w##l, aa=a##l, bb=b##l, cc=c##l, dd=d##l;          \
-      u32 *WW = (void*)&ww;                                     \
-      u32 *AA = (void*)&aa;                                     \
-      u32 *BB = (void*)&bb;                                     \
-      u32 *CC = (void*)&cc;                                     \
-      u32 *DD = (void*)&dd;                                     \
-      for (j=0; j<4; j++) {                                     \
-        printf ("%08x/%2i/%2i: %08x %08x %08x %08x\n",          \
-                WW[j], r, s,                                    \
-                AA[j], BB[j], CC[j], DD[j]);                    \
-      }                                                         \
-    }                                                           \
-    TTl = Fl(a,b,c,fun);                                        \
-    TTh = Fh(a,b,c,fun);                                        \
-    a##l = v32_rotate(a##l,r);                                  \
-    a##h = v32_rotate(a##h,r);                                  \
-    w##l  = v32_add(w##l, d##l);                                \
-    w##h  = v32_add(w##h, d##h);                                \
-    TTl = v32_add(TTl, w##l);                                   \
-    TTh = v32_add(TTh, w##h);                                   \
-    TTl = v32_rotate(TTl,s);                                    \
-    TTh = v32_rotate(TTh,s);                                    \
-    PERM(z,d,a);                                                \
-  } while(0)
-
-#define STEP_1(a,b,c,d,w,fun,r,s,z)             \
-  STEP_1_(a,b,c,d,w,fun,r,s,z)
-
-#define STEP_2_(a,b,c,d,w,fun,r,s)                               \
-  do {                                                          \
-    d##l = v32_add(d##l, TTl);                                  \
-    d##h = v32_add(d##h, TTh);                                  \
-  } while(0)
-
-#define STEP_2(a,b,c,d,w,fun,r,s)              \
-  STEP_2_(a,b,c,d,w,fun,r,s)
-  
-#define STEP(a,b,c,d,w1,w2,fun,r,s,z)           \
-  do {                                          \
-    register v32 TTl, TTh, Wl=w1, Wh=w2;        \
-    STEP_1(a,b,c,d,W,fun,r,s,z);                \
-    STEP_2(a,b,c,d,W,fun,r,s);                  \
-  } while(0);
-
-
-#define MSG_l(x) (2*(x))
-#define MSG_h(x) (2*(x)+1)
-
-#define MSG(w,hh,ll,u,z)                                \
-  do {                                                  \
-    int a = MSG_##u(hh);                                \
-    int b = MSG_##u(ll);                                \
-    w##l = v16_mergel(W[a], W[b]);                      \
-    w##l = V1632(v16_mul(V3216(w##l), code[z].v16));    \
-    w##h = v16_mergeh(W[a], W[b]);                      \
-    w##h = V1632(v16_mul(V3216(w##h), code[z].v16));    \
-  } while(0)
-  
-#define ROUND(h0,l0,u0,h1,l1,u1,h2,l2,u2,h3,l3,u3,        \
-              fun,r,s,t,u,z)                              \
-  do {                                                    \
-    register v32 W0l, W1l, W2l, W3l, TTl;                 \
-    register v32 W0h, W1h, W2h, W3h, TTh;                 \
-    MSG(W0,h0,l0,u0,z);                                   \
-    STEP_1(S(0), S(1), S(2), S(3), W0, fun, r, s, 0);     \
-    MSG(W1,h1,l1,u1,z);                                   \
-    STEP_2(S(0), S(1), S(2), S(3), W0, fun, r, s);        \
-    STEP_1(S(3), S(0), S(1), S(2), W1, fun, s, t, 1);     \
-    MSG(W2,h2,l2,u2,z);                                   \
-    STEP_2(S(3), S(0), S(1), S(2), W1, fun, s, t);        \
-    STEP_1(S(2), S(3), S(0), S(1), W2, fun, t, u, 2);     \
-    MSG(W3,h3,l3,u3,z);                                   \
-    STEP_2(S(2), S(3), S(0), S(1), W2, fun, t, u);        \
-    STEP_1(S(1), S(2), S(3), S(0), W3, fun, u, r, 3);     \
-    STEP_2(S(1), S(2), S(3), S(0), W3, fun, u, r);        \
-  } while(0)
-
-
-  /*
-   * 4 rounds with code 185
-   */
-#define PERM_START 0
-  ROUND(  2, 10, l,  3, 11, l,  0,  8, l,  1,  9, l, 0, 3,  23, 17, 27, 0);
-#undef PERM_START
-#define PERM_START 4
-  ROUND(  3, 11, h,  2, 10, h,  1,  9, h,  0,  8, h, 1, 3,  23, 17, 27, 0);
-#undef PERM_START
-#define PERM_START 1
-  ROUND(  7, 15, h,  5, 13, h,  6, 14, l,  4, 12, l, 0, 28, 19, 22, 7,  0);
-#undef PERM_START
-#define PERM_START 5
-  ROUND(  4, 12, h,  6, 14, h,  5, 13, l,  7, 15, l, 1, 28, 19, 22, 7,  0);
-#undef PERM_START
-
-  /*
-   * 4 rounds with code 233
-   */
-#define PERM_START 2
-  ROUND(  0,  4, h,  1,  5, l,  3,  7, h,  2,  6, l, 0, 29,  9, 15,  5, 1);
-#undef PERM_START
-#define PERM_START 6
-  ROUND(  3,  7, l,  2,  6, h,  0,  4, l,  1,  5, h, 1, 29,  9, 15,  5, 1);
-#undef PERM_START
-#define PERM_START 3
-  ROUND( 11, 15, l,  8, 12, l,  8, 12, h, 11, 15, h, 0,  4, 13, 10, 25, 1);
-#undef PERM_START
-#define PERM_START 0
-  ROUND(  9, 13, h, 10, 14, h, 10, 14, l,  9, 13, l, 1,  4, 13, 10, 25, 1);
-#undef PERM_START
-
-
-  /*
-   * 1 round as feed-forward
-   */
-#define PERM_START 4
-  STEP(S(0), S(1), S(2), S(3), S[0], S[1], 0,  4, 13, 0);
-  STEP(S(3), S(0), S(1), S(2), S[2], S[3], 0, 13, 10, 1);
-  STEP(S(2), S(3), S(0), S(1), S[4], S[5], 0, 10, 25, 2);
-  STEP(S(1), S(2), S(3), S(0), S[6], S[7], 0, 25,  4, 3);
-#undef PERM_START
-
-  S[0] = S0l;  S[1] = S0h;  S[2] = S1l;  S[3] = S1h;
-  S[4] = S2l;  S[5] = S2h;  S[6] = S3l;  S[7] = S3h;
-
-#undef ROUND
-#undef STEP
-#undef STEP_1
-#undef STEP_2
-}
-
-void SIMD_Compress(hashState_sd * state, const unsigned char *m, int final) {
-  if (state->hashbitlen <= 256) {
-    union cv Y[16];
-    short* y = (short*) Y[0].u16;
-
-#ifdef v16_broadcast
-    if (final == 2) {
-      fft128_msg_final(y, m);
-      rounds(state->A, m, y);
-    } else {
-      fft128_msg(y, m, final);
-      rounds(state->A, m, y);
-    }
-#else
-    fft128_msg(y, m, final);
-    rounds(state->A, m, y);
-#endif
-  } else {
-    union cv Y[32];
-    short* y = (short*) Y[0].u16;
-    
-    fft256_msg(y, m, final);
-    rounds512(state->A, m, y);
-  }
-}
-
-/* 
- * Give the FFT output in the regular order for consitancy checks
- */
-void fft128_natural(fft_t *x, unsigned char *a) {
-  union cv Y[16];
-  short* y = (short*) Y[0].u16;
-  int i;
-
-  fft128_msg(y, a, 0);
-
-  for(i=0; i<64; i++) {
-    x[2*i]   = y[i];
-    x[2*i+1] = y[i+64];
-  }
-}
-
-#endif // SSE2

algo/simd/vector.h

@@ -1,246 +0,0 @@
-#ifndef __VECTOR_H__
-#define __VECTOR_H__
-
-#include "compat.h"
-#include "simd-utils.h"
-
-/******************************* 
- * Using GCC vector extensions * 
- *******************************/
-
-//typedef unsigned char v16qi __attribute__ ((vector_size (16)));
-typedef char          v16qi __attribute__ ((vector_size (16)));
-typedef short          v8hi __attribute__ ((vector_size (16)));
-typedef int            v4si __attribute__ ((vector_size (16)));
-typedef float          v4sf __attribute__ ((vector_size (16)));
-typedef long long int  v2di __attribute__ ((vector_size (16)));
-
-typedef short          v4hi __attribute__ ((vector_size (8)));
-typedef unsigned char  v8qi __attribute__ ((vector_size (8)));
-
-typedef v16qi v8;
-typedef v8hi v16;
-typedef v4si v32;
-#define V16_SIZE 8
-
-union cv {
-  unsigned short u16[8];
-  v16 v16;
-};
-
-union cv8 {
-  unsigned char u8[16];
-  v8 v8;
-};
-
-union u32 {
-  u32 u[4];
-  v32 v;
-};
-
-#define V3216(x) ((v16) (x))
-#define V1632(x) ((v32) (x))
-#define  V168(x) ( (v8) (x))
-#define  V816(x) ((v16) (x))
-
-#if 0
-/* These instruction are shorter than the PAND/POR/... that GCC uses */
-
-#define vec_and(x,y)  ({v16 a = (v16) x; v16 b = (v16) y;  __builtin_ia32_andps ((v4sf) a, (v4sf) b);})
-#define vec_or(x,y)   ({v16 a = (v16) x; v16 b = (v16) y;  __builtin_ia32_orps ((v4sf) a, (v4sf) b);})
-#define vec_xor(x,y)  ({v16 a = (v16) x; v16 b = (v16) y;  __builtin_ia32_xorps ((v4sf) a, (v4sf) b);})
-#define vec_andn(x,y) ({v16 a = (v16) x; v16 b = (v16) y;  __builtin_ia32_andnps ((v4sf) a, (v4sf) b);})
-
-#define v16_and(x,y)  ((v16) vec_and ((x), (y)))
-#define v16_or(x,y)   ((v16) vec_or  ((x), (y)))
-#define v16_xor(x,y)  ((v16) vec_xor ((x), (y)))
-#define v16_andn(x,y) ((v16) vec_andn((x), (y)))
-
-#define v32_and(x,y)  ((v32) vec_and ((x), (y)))
-#define v32_or(x,y)   ((v32) vec_or  ((x), (y)))
-#define v32_xor(x,y)  ((v32) vec_xor ((x), (y)))
-#define v32_andn(x,y) ((v32) vec_andn((x), (y)))
-#endif
-
-#if defined(__SSE2__)
-
-#define vec_and(x,y) ((x)&(y))
-#define vec_or(x,y)  ((x)|(y))
-#define vec_xor(x,y) ((x)^(y))
-
-#define v16_and vec_and
-#define v16_or  vec_or
-#define v16_xor vec_xor
-
-#define v32_and vec_and
-#define v32_or  vec_or
-#define v32_xor vec_xor
-
-#define vec_andn(x,y) __builtin_ia32_pandn128 ((v2di) x, (v2di) y)
-#define v16_andn(x,y) ((v16) vec_andn(x,y))
-#define v32_andn(x,y) ((v32) vec_andn(x,y))
-
-#define v32_add(x,y) ((x)+(y))
-
-#define v16_add(x,y) ((x)+(y))
-#define v16_sub(x,y) ((x)-(y))
-#define v16_mul(x,y) ((x)*(y))
-#define v16_neg(x)   (-(x))
-#define v16_shift_l  __builtin_ia32_psllwi128
-#define v16_shift_r  __builtin_ia32_psrawi128
-#define v16_cmp      __builtin_ia32_pcmpgtw128
-
-#define v16_interleavel   __builtin_ia32_punpcklwd128
-#define v16_interleaveh   __builtin_ia32_punpckhwd128
-
-#define v16_mergel(a,b)   V1632(__builtin_ia32_punpcklwd128(a,b))
-#define v16_mergeh(a,b)   V1632(__builtin_ia32_punpckhwd128(a,b))
-
-#define v8_mergel(a,b) V816(__builtin_ia32_punpcklbw128(a,b))
-#define v8_mergeh(a,b) V816(__builtin_ia32_punpckhbw128(a,b))
-
-#define v32_shift_l  __builtin_ia32_pslldi128
-#define v32_shift_r  __builtin_ia32_psrldi128
-
-#define v32_rotate(x,n)                                 \
-  v32_or(v32_shift_l(x,n), v32_shift_r(x,32-(n)))
-
-#define v32_shuf __builtin_ia32_pshufd
-
-#define SHUFXOR_1 0xb1          /* 0b10110001 */
-#define SHUFXOR_2 0x4e          /* 0b01001110 */
-#define SHUFXOR_3 0x1b          /* 0b00011011 */
-
-#define CAT(x, y) x##y
-#define XCAT(x,y) CAT(x,y)
-
-#define v32_shufxor(x,s) v32_shuf(x,XCAT(SHUFXOR_,s))
-
-#define v32_bswap(x) (x)
-
-#define v16_broadcast(x) ({                     \
-      union u32 u;                              \
-      u32 xx = x;                               \
-      u.u[0] = xx | (xx << 16);                 \
-      V3216(v32_shuf(u.v,0)); })
-
-#define CV(x) {{x, x, x, x, x, x, x, x}}
-
-#elif defined(__aarch64__) && defined(__ARM_NEON)
-
-#define vec_and( x, y )    v128_and( x, y )
-#define vec_or(x,y)        v128_or( x, y )
-#define vec_xor(x,y)       v128_xor( x, y )
-
-#define v16_and v128_and
-#define v16_or  v128_or
-#define v16_xor v128_xor
-
-#define v32_and v128_and
-#define v32_or  v128_or
-#define v32_xor v128_xor
-
-#define vec_andn( x,y )   v128_andnot( x, y )
-#define v16_andn          vec_andn 
-#define v32_andn          vec_andn
-
-#define v32_add( x, y )   v128_add32( x, y )
-
-#define v16_add( x, y )        v128_add16( x, y )
-#define v16_sub( x, y )        v128_sub16( x, y )
-#define v16_mul( x, y )        v128_mul16( x, y )
-#define v16_neg(x)             v128_negate16( x )
-#define v16_shift_l( x, c )    v128_sl16
-#define v16_shift_r            v128_sr16
-#define v16_cmp                v128_cmpgt16
-
-#define v16_interleavel        v128_unpacklo16
-#define v16_interleaveh        v128_unpackhi16 
-
-#define v16_mergel(a,b)   V1632(__builtin_ia32_punpcklwd128(a,b))
-#define v16_mergeh(a,b)   V1632(__builtin_ia32_punpckhwd128(a,b))
-
-#define v8_mergel(a,b) V816(__builtin_ia32_punpcklbw128(a,b))
-#define v8_mergeh(a,b) V816(__builtin_ia32_punpckhbw128(a,b))
-
-#define v32_shift_l            v128_sl32
-#define v32_shift_r            v128_sr32
-
-#define v32_rotate(x,n)        v128_rol32
-
-#define v32_shuf __builtin_ia32_pshufd
-
-#define SHUFXOR_1 0xb1          /* 0b10110001 */
-#define SHUFXOR_2 0x4e          /* 0b01001110 */
-#define SHUFXOR_3 0x1b          /* 0b00011011 */
-
-#define CAT(x, y) x##y
-#define XCAT(x,y) CAT(x,y)
-
-#define v32_shufxor(x,s) v32_shuf(x,XCAT(SHUFXOR_,s))
-
-#define v32_bswap(x) (x)
-
-#define v16_broadcast(x) ({                     \
-      union u32 u;                              \
-      u32 xx = x;                               \
-      u.u[0] = xx | (xx << 16);                 \
-      V3216(v32_shuf(u.v,0)); })
-
-#define CV(x) {{x, x, x, x, x, x, x, x}}
-
-#else
-
-#error "I don't know how to vectorize on this architecture."
-
-#endif
-
-
-/* Twiddle tables */
-
-  static const union cv FFT64_Twiddle[] = {
-    {{1,    2,    4,    8,   16,   32,   64,  128}},
-    {{1,   60,    2,  120,    4,  -17,    8,  -34}},
-    {{1,  120,    8,  -68,   64,  -30,   -2,   17}},
-    {{1,   46,   60,  -67,    2,   92,  120,  123}},
-    {{1,   92,  -17,  -22,   32,  117,  -30,   67}},
-    {{1,  -67,  120,  -73,    8,  -22,  -68,  -70}},
-    {{1,  123,  -34,  -70,  128,   67,   17,   35}},
-  };
-
-
-  static const union cv FFT128_Twiddle[] =  {
-    {{  1, -118,   46,  -31,   60,  116,  -67,  -61}},
-    {{  2,   21,   92,  -62,  120,  -25,  123, -122}},
-    {{  4,   42,  -73, -124,  -17,  -50,  -11,   13}},
-    {{  8,   84,  111,    9,  -34, -100,  -22,   26}},
-    {{ 16,  -89,  -35,   18,  -68,   57,  -44,   52}},
-    {{ 32,   79,  -70,   36,  121,  114,  -88,  104}},
-    {{ 64,  -99,  117,   72,  -15,  -29,   81,  -49}},
-    {{128,   59,  -23, -113,  -30,  -58,  -95,  -98}},
-  };
-
-
-  static const union cv FFT256_Twiddle[] =  {
-    {{   1,   41, -118,   45,   46,   87,  -31,   14}}, 
-    {{  60, -110,  116, -127,  -67,   80,  -61,   69}}, 
-    {{   2,   82,   21,   90,   92,  -83,  -62,   28}}, 
-    {{ 120,   37,  -25,    3,  123,  -97, -122, -119}}, 
-    {{   4,  -93,   42,  -77,  -73,   91, -124,   56}}, 
-    {{ -17,   74,  -50,    6,  -11,   63,   13,   19}}, 
-    {{   8,   71,   84,  103,  111,  -75,    9,  112}}, 
-    {{ -34, -109, -100,   12,  -22,  126,   26,   38}}, 
-    {{  16, -115,  -89,  -51,  -35,  107,   18,  -33}}, 
-    {{ -68,   39,   57,   24,  -44,   -5,   52,   76}}, 
-    {{  32,   27,   79, -102,  -70,  -43,   36,  -66}}, 
-    {{ 121,   78,  114,   48,  -88,  -10,  104, -105}}, 
-    {{  64,   54,  -99,   53,  117,  -86,   72,  125}}, 
-    {{ -15, -101,  -29,   96,   81,  -20,  -49,   47}}, 
-    {{ 128,  108,   59,  106,  -23,   85, -113,   -7}}, 
-    {{ -30,   55,  -58,  -65,  -95,  -40,  -98,   94}}
-  };
-
-
-
-
-#endif

algo/x11/c11.c

@@ -13,11 +13,7 @@
 #include "algo/skein/sph_skein.h"
 #include "algo/shavite/sph_shavite.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #include "algo/luffa/luffa_for_sse2.h"
 #if defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"
@@ -43,11 +39,7 @@ typedef struct {
    hashState_luffa         luffa;
    cubehashParam           cube;
    sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+   simd512_context         simd;
 } c11_ctx_holder;
 
 c11_ctx_holder c11_ctx __attribute__ ((aligned (64)));
@@ -69,11 +61,6 @@ void init_c11_ctx()
    init_luffa( &c11_ctx.luffa, 512 );
    cubehashInit( &c11_ctx.cube, 512, 16, 32 );
    sph_shavite512_init( &c11_ctx.shavite );
-#if defined(__aarch64__)
-   sph_simd512_init( &c11_ctx.simd );
-#else
-   init_sd( &c11_ctx.simd, 512 );
-#endif
 }
 
 void c11_hash( void *output, const void *input )
@@ -105,41 +92,35 @@ void c11_hash( void *output, const void *input )
     sph_skein512( &ctx.skein, (const void*) hash, 64 );
     sph_skein512_close( &ctx.skein, hash );
 
-     update_and_final_luffa( &ctx.luffa, hash, hash, 64 );
+    update_and_final_luffa( &ctx.luffa, hash, hash, 64 );
 
-     cubehashUpdateDigest( &ctx.cube, hash, hash, 64 );
+    cubehashUpdateDigest( &ctx.cube, hash, hash, 64 );
 
-     sph_shavite512( &ctx.shavite, hash, 64);
-     sph_shavite512_close( &ctx.shavite, hash);
+    sph_shavite512( &ctx.shavite, hash, 64);
+    sph_shavite512_close( &ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                                   (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
-     update_final_echo ( &ctx.echo, (BitSequence *)hash,
-                         (const BitSequence *)hash, 512 );
+    update_final_echo ( &ctx.echo, (BitSequence *)hash,
+                        (const BitSequence *)hash, 512 );
 #else
-     sph_echo512( &ctx.echo, hash, 64 );
-     sph_echo512_close( &ctx.echo, hash );
+    sph_echo512( &ctx.echo, hash, 64 );
+    sph_echo512_close( &ctx.echo, hash );
 #endif
 
-        memcpy(output, hash, 32);
+    memcpy(output, hash, 32);
 }
 
 int scanhash_c11( struct work *work, uint32_t max_nonce,
                   uint64_t *hashes_done, struct thr_info *mythr )
 {
-        uint32_t endiandata[20] __attribute__((aligned(64)));
-        uint32_t hash[8] __attribute__((aligned(64)));
-        uint32_t *pdata = work->data;
-        uint32_t *ptarget = work->target;
+   uint32_t endiandata[20] __attribute__((aligned(64)));
+   uint32_t hash[8] __attribute__((aligned(64)));
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
 	const uint32_t first_nonce = pdata[19];
-        const uint32_t Htarg = ptarget[7];
+   const uint32_t Htarg = ptarget[7];
 	uint32_t nonce = first_nonce;
    int thr_id = mythr->id;
 	volatile uint8_t *restart = &(work_restart[thr_id].restart);

algo/x11/timetravel10.c

@@ -13,17 +13,13 @@
 #include "algo/skein/sph_skein.h"
 #include "algo/cubehash/cubehash_sse2.h"
 #include "algo/shavite/sph_shavite.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #ifdef __AES__
   #include "algo/groestl/aes_ni/hash-groestl.h"
 #else
   #include "algo/groestl/sph_groestl.h"
 #endif
-  #include "algo/luffa/luffa_for_sse2.h"
+#include "algo/luffa/luffa_for_sse2.h"
 
 static __thread uint32_t s_ntime = UINT32_MAX;
 static __thread int permutation[TT10_FUNC_COUNT] = { 0 };
@@ -37,11 +33,7 @@ typedef struct {
         hashState_luffa         luffa;
         cubehashParam           cube;
         sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+        simd512_context         simd;
 #ifdef __AES__
         hashState_groestl       groestl;
 #else
@@ -62,11 +54,6 @@ void init_tt10_ctx()
         init_luffa( &tt10_ctx.luffa, 512 );
         cubehashInit( &tt10_ctx.cube, 512, 16, 32 );
         sph_shavite512_init( &tt10_ctx.shavite );
-#if defined(__aarch64__)
-   sph_simd512_init( &tt10_ctx.simd );
-#else
-   init_sd( &tt10_ctx.simd, 512 );
-#endif
 #ifdef __AES__
         init_groestl( &tt10_ctx.groestl, 64 );
 #else
@@ -222,27 +209,7 @@ void timetravel10_hash(void *output, const void *input)
         }
         break;
      case 9:
-        if ( i == 0 )
-        {
-           memcpy( &ctx.simd, &tt10_mid.simd, sizeof tt10_mid.simd );
-#if defined(__aarch64__)
-           sph_simd512(&ctx.simd, (const void*) input + midlen, tail );
-           sph_simd512_close(&ctx.simd, hash);
-#else
-           update_final_sd( &ctx.simd, (BitSequence *)hashB,
-                            (const BitSequence *)input + midlen, tail*8 );
-#endif
-        }
-        else
-        {
-#if defined(__aarch64__)
-           sph_simd512(&ctx.simd, (const void*) hash, 64);
-           sph_simd512_close(&ctx.simd, hash);
-#else
-           update_sd( &ctx.simd, (const BitSequence *)hashA, dataLen*8 );
-           final_sd( &ctx.simd, (BitSequence *)hashB );
-#endif
-        }
+        simd512_ctx( &ctx.simd, hashB, hashA, dataLen );
         break;
      default:
 	break;
@@ -325,15 +292,6 @@ int scanhash_timetravel10( struct work *work, uint32_t max_nonce,
            memcpy( &tt10_mid.shavite, &tt10_ctx.shavite, sizeof(tt10_mid.shavite ) );
            sph_shavite512( &tt10_mid.shavite, endiandata, 64 );
            break;
-        case 9:
-           memcpy( &tt10_mid.simd, &tt10_ctx.simd, sizeof(tt10_mid.simd ) );
-#if defined(__aarch64__)
-           sph_simd512( &tt10_mid.simd, (const void*) endiandata, 64 );
-           sph_simd512_close( &tt10_mid.simd, hash);
-#else
-           update_sd( &tt10_mid.simd, (const BitSequence *)endiandata, 512 );
-#endif
-           break;
         default:
            break;
       }

algo/x11/x11.c

@@ -22,12 +22,7 @@
   #include "algo/echo/sph_echo.h"
 #endif
 #include "algo/luffa/luffa_for_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
-
+#include "algo/simd/simd-hash-2way.h"
 
 typedef struct {
    sph_blake512_context blake;
@@ -45,11 +40,7 @@ typedef struct {
    hashState_luffa         luffa;
    cubehashParam           cube;
    sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+   simd512_context         simd;
 } x11_ctx_holder;
 
 x11_ctx_holder x11_ctx;
@@ -71,11 +62,6 @@ void init_x11_ctx()
    init_luffa( &x11_ctx.luffa, 512 );
    cubehashInit( &x11_ctx.cube, 512, 16, 32 );
    sph_shavite512_init( &x11_ctx.shavite );
-#if defined(__aarch64__)
-   sph_simd512_init( &x11_ctx.simd );
-#else
-   init_sd( &x11_ctx.simd, 512 );
-#endif
 }
 
 void x11_hash( void *state, const void *input )
@@ -118,13 +104,7 @@ void x11_hash( void *state, const void *input )
     sph_shavite512( &ctx.shavite, hash, 64 );
     sph_shavite512_close( &ctx.shavite, hash );
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                       (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
     update_final_echo ( &ctx.echo, (BitSequence *)hash,

algo/x11/x11evo.c

@@ -20,11 +20,7 @@
   #include "algo/echo/sph_echo.h"
 #endif
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #include "algo/luffa/luffa_for_sse2.h"
 
 typedef struct {
@@ -37,11 +33,7 @@ typedef struct {
 #endif
     hashState_luffa         luffa;
     cubehashParam           cube;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+   simd512_context         simd;
     sph_blake512_context    blake;
     sph_bmw512_context      bmw;
     sph_skein512_context    skein;
@@ -63,11 +55,6 @@ void init_x11evo_ctx()
 #endif
      init_luffa( &x11evo_ctx.luffa, 512 );
      cubehashInit( &x11evo_ctx.cube, 512, 16, 32 );
-#if defined(__aarch64__)
-     sph_simd512_init( &x11evo_ctx.simd );
-#else
-     init_sd( &x11evo_ctx.simd, 512 );
-#endif
      sph_blake512_init( &x11evo_ctx.blake );
      sph_bmw512_init( &x11evo_ctx.bmw );
      sph_skein512_init( &x11evo_ctx.skein );
@@ -146,12 +133,7 @@ void x11evo_hash( void *state, const void *input )
 	      sph_shavite512_close( &ctx.shavite, (char*)hash );
 	      break;
 	    case 9:
-#if defined(__aarch64__)
-         sph_simd512(&ctx.simd, (const void*) hash, 64);
-         sph_simd512_close(&ctx.simd, hash);
-#else
-         update_final_sd( &ctx.simd, (char*)hash, (const char*)hash, 512 );
-#endif
+         simd512_ctx( &ctx.simd, hash, hash, 64 );
     break;
 	    case 10:
 #ifdef __AES__

algo/x11/x11gost.c

@@ -17,12 +17,7 @@
 #include "algo/shavite/sph_shavite.h"
 #include "algo/luffa/luffa_for_sse2.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-#include "algo/simd/sph_simd.h"
-#else
-#include "algo/simd/nist.h"
-#endif
-
+#include "algo/simd/simd-hash-2way.h"
 #if defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"
   #include "algo/groestl/aes_ni/hash-groestl.h"
@@ -47,11 +42,7 @@ typedef struct {
    hashState_luffa         luffa;
    cubehashParam           cube;
    sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-  sph_simd512_context     simd;
-#else
-  hashState_sd            simd;
-#endif
+   simd512_context         simd;
    sph_gost512_context     gost;
 } x11gost_ctx_holder;
 
@@ -75,11 +66,6 @@ void init_x11gost_ctx()
    sph_shavite512_init( &x11gost_ctx.shavite );
    init_luffa( &x11gost_ctx.luffa, 512 );
    cubehashInit( &x11gost_ctx.cube, 512, 16, 32 );
-#if defined(__aarch64__)
-    sph_simd512_init(&x11gost_ctx.simd);
-#else
-    init_sd( &x11gost_ctx.simd, 512 );
-#endif
 }
 
 void x11gost_hash(void *output, const void *input)
@@ -123,13 +109,7 @@ void x11gost_hash(void *output, const void *input)
     sph_shavite512( &ctx.shavite, hash, 64 );
     sph_shavite512_close( &ctx.shavite, hash );
 
-#if defined(__aarch64__)
-    sph_simd512 (&ctx.simd, hash, 64); 
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                                   (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
      update_final_echo ( &ctx.echo, (BitSequence *)hash,

algo/x12/x12.c

@@ -17,11 +17,7 @@
 #include "algo/echo/sph_echo.h"
 #include "algo/hamsi/sph_hamsi.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #if defined(__AES__)
   #include "algo/groestl/aes_ni/hash-groestl.h"
   #include "algo/echo/aes_ni/hash_api.h"
@@ -44,11 +40,7 @@ typedef struct {
    hashState_luffa         luffa;
    cubehashParam            cubehash;
    sph_shavite512_context   shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+   simd512_context         simd;
    sph_hamsi512_context     hamsi;
 } x12_ctx_holder;
 
@@ -68,14 +60,9 @@ void init_x12_ctx()
         sph_groestl512_init(&x12_ctx.groestl);
         sph_echo512_init(&x12_ctx.echo);
 #endif
-   init_luffa( &x12_ctx.luffa, 512 );
+        init_luffa( &x12_ctx.luffa, 512 );
         cubehashInit( &x12_ctx.cubehash, 512, 16, 32 );
         sph_shavite512_init( &x12_ctx.shavite );
-#if defined(__aarch64__)
-   sph_simd512_init( &x12_ctx.simd );
-#else
-   init_sd( &x12_ctx.simd, 512 );
-#endif
         sph_hamsi512_init( &x12_ctx.hamsi );
 };
 
@@ -101,13 +88,7 @@ void x12hash(void *output, const void *input)
    sph_shavite512( &ctx.shavite, hash, 64);
    sph_shavite512_close( &ctx.shavite, hashB);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hashB, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_sd( &ctx.simd, (const BitSequence *)hash, 512 );
-    final_sd( &ctx.simd, (BitSequence *)hash );
-#endif
+   simd512_ctx( &ctx.simd, hash, hashB, 64 );
 
 #if defined(__AES__)
    update_final_echo ( &ctx.echo, (BitSequence *)hashB,

algo/x13/x13.c

@@ -15,11 +15,7 @@
 #include "algo/hamsi/sph_hamsi.h"
 #include "algo/luffa/luffa_for_sse2.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-#include "algo/simd/sph_simd.h"
-#else
-#include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #if defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"
   #include "algo/groestl/aes_ni/hash-groestl.h"
@@ -48,11 +44,7 @@ typedef struct {
    hashState_luffa         luffa;
    cubehashParam           cubehash;
    sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-  sph_simd512_context     simd;
-#else
-  hashState_sd            simd;
-#endif
+   simd512_context         simd;
    sph_hamsi512_context    hamsi;
 } x13_ctx_holder;
 
@@ -77,11 +69,6 @@ void init_x13_ctx()
    init_luffa( &x13_ctx.luffa, 512 );
    cubehashInit( &x13_ctx.cubehash, 512, 16, 32 );
    sph_shavite512_init( &x13_ctx.shavite );
-#if defined(__aarch64__)
-    sph_simd512_init(&x13_ctx.simd);
-#else
-    init_sd( &x13_ctx.simd, 512 );
-#endif
    sph_hamsi512_init( &x13_ctx.hamsi );
 };
 
@@ -121,13 +108,7 @@ void x13hash(void *output, const void *input)
     sph_shavite512( &ctx.shavite, hash, 64);
     sph_shavite512_close( &ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                                   (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
     update_final_echo ( &ctx.echo, (BitSequence *)hash,

algo/x13/x13bcd.c

@@ -15,11 +15,7 @@
 #include "algo/shavite/sph_shavite.h"
 #include "algo/hamsi/sph_hamsi.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #if defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"
   #include "algo/groestl/aes_ni/hash-groestl.h"
@@ -47,11 +43,7 @@ typedef struct {
    sph_skein512_context    skein;
    cubehashParam           cube;
    sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+   simd512_context         simd;
    sph_hamsi512_context    hamsi;
    sm3_ctx_t               sm3;
 } x13bcd_ctx_holder;
@@ -76,11 +68,6 @@ void init_x13bcd_ctx()
    sph_keccak512_init( &x13bcd_ctx.keccak );
    cubehashInit( &x13bcd_ctx.cube,512,16,32 );
    sph_shavite512_init( &x13bcd_ctx.shavite );
-#if defined(__aarch64__)
-   sph_simd512_init( &x13bcd_ctx.simd );
-#else
-   init_sd( &x13bcd_ctx.simd, 512 );
-#endif
    sm3_init( &x13bcd_ctx.sm3 );
    sph_hamsi512_init( &x13bcd_ctx.hamsi );
 };
@@ -127,13 +114,7 @@ void x13bcd_hash(void *output, const void *input)
     sph_shavite512( &ctx.shavite, hash, 64);
     sph_shavite512_close( &ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                                   (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
     update_final_echo ( &ctx.echo, (BitSequence *)hash,

algo/x13/x13sm3.c

@@ -17,11 +17,7 @@
 #include "algo/fugue/sph_fugue.h"
 #include "algo/luffa/luffa_for_sse2.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #if defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"
   #include "algo/groestl/aes_ni/hash-groestl.h"
@@ -46,11 +42,7 @@ typedef struct {
    hashState_luffa         luffa;
    cubehashParam           cube;
    sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+   simd512_context         simd;
    sm3_ctx_t               sm3;
    sph_hamsi512_context    hamsi;
    sph_fugue512_context    fugue;
@@ -75,11 +67,6 @@ void init_x13sm3_ctx()
    init_luffa( &hsr_ctx.luffa,512 );
    cubehashInit( &hsr_ctx.cube,512,16,32 );
    sph_shavite512_init( &hsr_ctx.shavite );
-#if defined(__aarch64__)
-   sph_simd512_init( &hsr_ctx.simd );
-#else
-   init_sd( &hsr_ctx.simd,512 );
-#endif
    sm3_init( &hsr_ctx.sm3 );
    sph_hamsi512_init( &hsr_ctx.hamsi );
    sph_fugue512_init( &hsr_ctx.fugue );
@@ -123,13 +110,7 @@ void x13sm3_hash(void *output, const void *input)
     sph_shavite512( &ctx.shavite, hash, 64);
         sph_shavite512_close( &ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                                   (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );
 
         //11---echo---
 #ifdef __AES__

algo/x14/x14.c

@@ -15,11 +15,7 @@
 #include "algo/hamsi/sph_hamsi.h"
 #include "algo/shabal/sph_shabal.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #if defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"
   #include "algo/groestl/aes_ni/hash-groestl.h"
@@ -49,11 +45,7 @@ typedef struct {
    hashState_luffa         luffa;
    cubehashParam           cube;
    sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+   simd512_context         simd;
    sph_hamsi512_context    hamsi;
    sph_shabal512_context   shabal;
 } x14_ctx_holder;
@@ -79,11 +71,6 @@ void init_x14_ctx()
    init_luffa( &x14_ctx.luffa,512 );
    cubehashInit( &x14_ctx.cube,512,16,32 );
    sph_shavite512_init( &x14_ctx.shavite );
-#if defined(__aarch64__)
-   sph_simd512_init( &x14_ctx.simd );
-#else
-   init_sd( &x14_ctx.simd, 512 );
-#endif
    sph_hamsi512_init( &x14_ctx.hamsi );
    sph_shabal512_init( &x14_ctx.shabal );
 };
@@ -124,13 +111,7 @@ void x14hash(void *output, const void *input)
     sph_shavite512( &ctx.shavite, hash, 64);
     sph_shavite512_close( &ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                                   (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
     update_final_echo ( &ctx.echo, (BitSequence *)hash,

algo/x15/x15.c

@@ -17,12 +17,7 @@
 #include "algo/shabal/sph_shabal.h"
 #include "algo/whirlpool/sph_whirlpool.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
-
+#include "algo/simd/simd-hash-2way.h"
 #if defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"
   #include "algo/groestl/aes_ni/hash-groestl.h"
@@ -52,11 +47,7 @@ typedef struct {
    hashState_luffa         luffa;
    cubehashParam           cubehash;
    sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
+   simd512_context         simd;
    sph_hamsi512_context    hamsi;
    sph_shabal512_context   shabal;
    sph_whirlpool_context   whirlpool;
@@ -83,11 +74,6 @@ void init_x15_ctx()
    init_luffa( &x15_ctx.luffa,512 );
    cubehashInit( &x15_ctx.cubehash, 512, 16, 32 );
    sph_shavite512_init( &x15_ctx.shavite );
-#if defined(__aarch64__)
-   sph_simd512_init( &x15_ctx.simd );
-#else
-   init_sd( &x15_ctx.simd, 512 );
-#endif
    sph_hamsi512_init( &x15_ctx.hamsi );
    sph_shabal512_init( &x15_ctx.shabal );
    sph_whirlpool_init( &x15_ctx.whirlpool );
@@ -131,13 +117,7 @@ void x15hash(void *output, const void *input)
     sph_shavite512( &ctx.shavite, hash, 64);
     sph_shavite512_close( &ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                                   (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
     update_final_echo ( &ctx.echo, (BitSequence *)hash,

algo/x16/hex.c

@@ -236,7 +236,7 @@ int scanhash_hex( struct work *work, uint32_t max_nonce,
    do
    {
       edata[19] = nonce;
-      if ( hex_hash( hash32, edata, thr_id ) );
+      if ( hex_hash( hash32, edata, thr_id ) )
       if ( unlikely( valid_hash( hash32, ptarget ) && !bench ) )
       {
          be32enc( &pdata[19], nonce );

algo/x16/x16r-4way.c

@@ -526,7 +526,7 @@ int scanhash_x16r_8way( struct work *work, uint32_t max_nonce,
                              n+3, 0, n+2, 0, n+1, 0, n,   0 ), *noncev );
    do
    {
-      if( x16r_8way_hash( hash, vdata, thr_id ) );
+      if ( x16r_8way_hash( hash, vdata, thr_id ) )
       for ( int i = 0; i < 8; i++ )
       if ( unlikely( valid_hash( hash + (i<<3), ptarget ) && !bench ) )
       {
@@ -952,7 +952,7 @@ int scanhash_x16r_4way( struct work *work, uint32_t max_nonce,
                    _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
    do
    {
-      if ( x16r_4way_hash( hash, vdata, thr_id ) );
+      if ( x16r_4way_hash( hash, vdata, thr_id ) )
       for ( int i = 0; i < 4; i++ )
       if ( unlikely( valid_hash( hash + (i<<3), ptarget ) && !bench ) )
       {
@@ -1353,7 +1353,7 @@ int scanhash_x16r_2x64( struct work *work, uint32_t max_nonce,
    *noncev = v128_intrlv_blend_32( v128_set32( n+1, 0, n, 0 ), *noncev );
    do
    {
-      if ( x16r_2x64_hash( hash, vdata, thr_id ) );
+      if ( x16r_2x64_hash( hash, vdata, thr_id ) )
       for ( int i = 0; i < 2; i++ )
       if ( unlikely( valid_hash( hash + (i<<3), ptarget ) && !bench ) )
       {

algo/x16/x16r-gate.h

@@ -15,7 +15,6 @@
 #include "algo/luffa/luffa_for_sse2.h"
 #include "algo/cubehash/cubehash_sse2.h"
 #include "algo/simd/sph_simd.h"
-#include "algo/simd/nist.h"
 #include "algo/echo/sph_echo.h"
 #include "algo/hamsi/sph_hamsi.h"
 #include "algo/fugue/sph_fugue.h"

algo/x16/x20r.c

@@ -137,7 +137,7 @@ int scanhash_x20r_8x64( struct work *work, uint32_t max_nonce,
                              n+3, 0, n+2, 0, n+1, 0, n,   0 ), *noncev );
    do
    {
-      if( x20r_8x64_hash( hash, vdata, thr_id ) );
+      if ( x20r_8x64_hash( hash, vdata, thr_id ) )
       for ( int i = 0; i < 8; i++ )
       if ( unlikely( valid_hash( hash + (i<<3), ptarget ) && !bench ) )
       {
@@ -205,7 +205,7 @@ int scanhash_x20r_4x64( struct work *work, uint32_t max_nonce,
                    _mm256_set_epi32( n+3, 0, n+2, 0, n+1, 0, n, 0 ), *noncev );
    do
    {
-      if ( x20r_4x64_hash( hash, vdata, thr_id ) );
+      if ( x20r_4x64_hash( hash, vdata, thr_id ) )
       for ( int i = 0; i < 4; i++ )
       if ( unlikely( valid_hash( hash + (i<<3), ptarget ) && !bench ) )
       {
@@ -269,7 +269,7 @@ int scanhash_x20r_2x64( struct work *work, uint32_t max_nonce,
    *noncev = v128_intrlv_blend_32( v128_set32( n+1, 0, n, 0 ), *noncev );
    do
    {
-      if ( x20r_2x64_hash( hash, vdata, thr_id ) );
+      if ( x20r_2x64_hash( hash, vdata, thr_id ) )
       for ( int i = 0; i < 2; i++ )
       if ( unlikely( valid_hash( hash + (i<<3), ptarget ) && !bench ) )
       {

algo/x17/sonoa.c

@@ -18,11 +18,7 @@
 #include "algo/whirlpool/sph_whirlpool.h"
 #include "algo/haval/sph-haval.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #include "algo/sha/sph_sha2.h"
 #if defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"
@@ -53,11 +49,7 @@ typedef struct {
         hashState_luffa         luffa;
         cubehashParam           cubehash;
         sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-        sph_simd512_context     simd;
-#else
-        hashState_sd            simd;
-#endif
+        simd512_context         simd;
         sph_hamsi512_context    hamsi;
         sph_shabal512_context   shabal;
         sph_whirlpool_context   whirlpool;
@@ -86,11 +78,6 @@ void init_sonoa_ctx()
         init_luffa( &sonoa_ctx.luffa, 512 );
         cubehashInit( &sonoa_ctx.cubehash, 512, 16, 32 );
         sph_shavite512_init( &sonoa_ctx.shavite );
-#if defined(__aarch64__)
-        sph_simd512_init( &sonoa_ctx.simd );
-#else
-        init_sd( &sonoa_ctx.simd, 512 );
-#endif
         sph_hamsi512_init( &sonoa_ctx.hamsi );
         sph_shabal512_init( &sonoa_ctx.shabal );
         sph_whirlpool_init( &sonoa_ctx.whirlpool );
@@ -134,13 +121,7 @@ int sonoa_hash( void *state, const void *input, int thr_id )
 	sph_shavite512(&ctx.shavite, hash, 64);
 	sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                       (const BitSequence *)hash, 512 );
-#endif
+   simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
    update_final_echo ( &ctx.echo, (BitSequence *)hash,
@@ -189,13 +170,7 @@ int sonoa_hash( void *state, const void *input, int thr_id )
    sph_shavite512(&ctx.shavite, hash, 64);
    sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-   sph_simd512(&ctx.simd, (const void*) hash, 64);
-   sph_simd512_close(&ctx.simd, hash);
-#else
-   update_final_sd( &ctx.simd, (BitSequence *)hash,
-                       (const BitSequence *)hash, 512 );
-#endif
+   simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
    init_echo( &ctx.echo, 512 );
@@ -249,13 +224,7 @@ int sonoa_hash( void *state, const void *input, int thr_id )
    sph_shavite512(&ctx.shavite, hash, 64);
    sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-   sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-   update_final_sd( &ctx.simd, (BitSequence *)hash,
-                       (const BitSequence *)hash, 512 );
-#endif
+   simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
    init_echo( &ctx.echo, 512 );
@@ -318,13 +287,7 @@ int sonoa_hash( void *state, const void *input, int thr_id )
    sph_shavite512(&ctx.shavite, hash, 64);
    sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                       (const BitSequence *)hash, 512 );
-#endif
+   simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
    init_echo( &ctx.echo, 512 );
@@ -410,13 +373,7 @@ int sonoa_hash( void *state, const void *input, int thr_id )
    sph_shavite512(&ctx.shavite, hash, 64);
    sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512_init( &ctx.simd );
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    simd_full( &ctx.simd, hash, hash, 512 );
-#endif
+   simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
    init_echo( &ctx.echo, 512 );
@@ -483,13 +440,7 @@ int sonoa_hash( void *state, const void *input, int thr_id )
    sph_shavite512(&ctx.shavite, hash, 64);
    sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512_init( &ctx.simd );
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    simd_full( &ctx.simd, hash, hash, 512 );
-#endif
+   simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
    init_echo( &ctx.echo, 512 );
@@ -527,7 +478,6 @@ int sonoa_hash( void *state, const void *input, int thr_id )
    sph_whirlpool_close(&ctx.whirlpool, hash);
 
    if ( work_restart[thr_id].restart ) return 0;
-//
 
    sph_bmw512_init( &ctx.bmw);
    sph_bmw512(&ctx.bmw, hash, 64);
@@ -565,13 +515,7 @@ int sonoa_hash( void *state, const void *input, int thr_id )
    sph_shavite512(&ctx.shavite, hash, 64);
    sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512_init( &ctx.simd );
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    simd_full( &ctx.simd, hash, hash, 512 );
-#endif
+   simd512_ctx( &ctx.simd, hash, hash, 64 );
 
 #if defined(__AES__)
    init_echo( &ctx.echo, 512 );

algo/x17/x17.c

@@ -18,11 +18,7 @@
 #include "algo/whirlpool/sph_whirlpool.h"
 #include "algo/haval/sph-haval.h"
 #include "algo/cubehash/cubehash_sse2.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #include "algo/sha/sph_sha2.h"
 #if defined(__AES__)
   #include "algo/fugue/fugue-aesni.h"
@@ -34,7 +30,7 @@
   #include "algo/fugue/sph_fugue.h"
 #endif
 #include "algo/blake/sph_blake.h"
-#include "algo/cubehash/sph_cubehash.h"
+//#include "algo/cubehash/sph_cubehash.h"
 #include "algo/luffa/sph_luffa.h"
 
 
@@ -63,17 +59,9 @@ union _x17_context_overlay
 #else
         hashState_luffa         luffa;
 #endif
-//#if defined(__aarch64__)
-//        sph_cubehash512_context    cube;
-//#else
         cubehashParam           cube;
-//#endif
         sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-        sph_simd512_context     simd;
-#else
-        hashState_sd            simd;
-#endif
+        simd512_context         simd;
         sph_hamsi512_context    hamsi;
         sph_shabal512_context   shabal;
         sph_whirlpool_context   whirlpool;
@@ -127,26 +115,13 @@ int x17_hash(void *output, const void *input, int thr_id )
     luffa_full( &ctx.luffa, hash, 512, hash, 64 );
 #endif
 
-//#if defined(__aarch64__)
-//    sph_cubehash512_init(&ctx.cube);
-//    sph_cubehash512(&ctx.cube, (const void*) hash, 64);
-//    sph_cubehash512_close(&ctx.cube, hash);
-//#else
     cubehash_full( &ctx.cube, hash, 512, hash, 64 );
-//#endif
 
     sph_shavite512_init( &ctx.shavite );
     sph_shavite512( &ctx.shavite, hash, 64);
     sph_shavite512_close( &ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512_init( &ctx.simd );
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-    simd_full( &ctx.simd, (BitSequence *)hash,
-               (const BitSequence *)hash, 512 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, 64 );        
 
 #if defined(__AES__)
     echo_full( &ctx.echo, (BitSequence *)hash, 512,

algo/x17/xevan.c

@@ -17,11 +17,7 @@
 #include "algo/shabal/sph_shabal.h"
 #include "algo/whirlpool/sph_whirlpool.h"
 #include "algo/haval/sph-haval.h"
-#if defined(__aarch64__)
-  #include "algo/simd/sph_simd.h"
-#else
-  #include "algo/simd/nist.h"
-#endif
+#include "algo/simd/simd-hash-2way.h"
 #include "algo/cubehash/cubehash_sse2.h"
 #include "algo/sha/sph_sha2.h"
 #if defined(__AES__)
@@ -45,11 +41,7 @@ typedef struct {
         hashState_luffa         luffa;
         cubehashParam           cubehash;
         sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-        sph_simd512_context     simd;
-#else
-        hashState_sd            simd;
-#endif
+        simd512_context         simd;
         sph_hamsi512_context    hamsi;
         sph_shabal512_context   shabal;
         sph_whirlpool_context   whirlpool;
@@ -78,11 +70,6 @@ void init_xevan_ctx()
         init_luffa( &xevan_ctx.luffa, 512 );
         cubehashInit( &xevan_ctx.cubehash, 512, 16, 32 );
         sph_shavite512_init( &xevan_ctx.shavite );
-#if defined(__aarch64__)
-        sph_simd512_init( &xevan_ctx.simd );
-#else
-        init_sd( &xevan_ctx.simd, 512 );
-#endif
         sph_hamsi512_init( &xevan_ctx.hamsi );
         sph_shabal512_init( &xevan_ctx.shabal );
         sph_whirlpool_init( &xevan_ctx.whirlpool );
@@ -137,13 +124,7 @@ int xevan_hash(void *output, const void *input, int thr_id )
    sph_shavite512(&ctx.shavite, hash, dataLen);
    sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512( &ctx.simd, (const void*) hash, dataLen );
-    sph_simd512_close( &ctx.simd, hash );
-#else
-    update_final_sd( &ctx.simd, (BitSequence *)hash,
-                         (const BitSequence *)hash, dataLen*8 );
-#endif
+   simd512_ctx( &ctx.simd, hash, hash, dataLen );
 
 #if defined(__AES__)
    update_final_echo( &ctx.echo, (BitSequence *) hash,
@@ -210,13 +191,14 @@ int xevan_hash(void *output, const void *input, int thr_id )
    sph_shavite512(&ctx.shavite, hash, dataLen);
    sph_shavite512_close(&ctx.shavite, hash);
 
-#if defined(__aarch64__)
-    sph_simd512(&ctx.simd, (const void*) hash, 64);
-    sph_simd512_close(&ctx.simd, hash);
-#else
-   update_final_sd( &ctx.simd, (BitSequence *)hash,
-                         (const BitSequence *)hash, dataLen*8 );
-#endif
+    simd512_ctx( &ctx.simd, hash, hash, dataLen );
+//#if defined(__aarch64__)
+//    sph_simd512(&ctx.simd, (const void*) hash, 64);
+//    sph_simd512_close(&ctx.simd, hash);
+//#else
+//   update_final_sd( &ctx.simd, (BitSequence *)hash,
+//                         (const BitSequence *)hash, dataLen*8 );
+//#endif
 
 #if defined(__AES__)
    update_final_echo( &ctx.echo, (BitSequence *) hash,

algo/x22/x25x-4way.c

@@ -18,7 +18,6 @@
 #include "algo/cubehash/cube-hash-2way.h"
 #include "algo/shavite/shavite-hash-2way.h"
 #include "algo/shavite/sph_shavite.h"
-#include "algo/simd/nist.h"
 #include "algo/simd/simd-hash-2way.h"
 #include "algo/fugue/fugue-aesni.h"
 #include "algo/whirlpool/sph_whirlpool.h"

arm-build.sh

@@ -1,14 +0,0 @@
-#!/bin/bash
-
-# Linux build
-
-make distclean || echo clean
-
-rm -f config.status
-./autogen.sh || echo done
-
-CFLAGS="-O3 -march=native -Wall -flax-vector-conversions" ./configure  --with-curl 
-
-make -j $(nproc)
-
-strip -s cpuminer

armbuild-all.sh

@@ -15,51 +15,43 @@ rm -f config.status
 ./autogen.sh || echo done
 CFLAGS="-O3 -march=armv9-a+crypto+sha3 -Wall -flax-vector-conversions" ./configure  --with-curl
 make -j $(nproc)
-strip -s cpuminer
 mv cpuminer cpuminer-armv9-crypto-sha3
 
 make clean || echo clean
 CFLAGS="-O3 -march=armv9-a+crypto -Wall -flax-vector-conversions" ./configure  --with-curl
 make -j $(nproc)
-strip -s cpuminer
 mv cpuminer cpuminer-armv9-crypto
 
 make clean || echo clean
 CFLAGS="-O3 -march=armv9-a -Wall -flax-vector-conversions" ./configure  --with-curl
 make -j $(nproc)
-strip -s cpuminer
 mv cpuminer cpuminer-armv9
 
 # SVE2 available in armv8.5
 make clean || echo clean
 CFLAGS="-O3 -march=armv8.5-a+crypto+sha3+sve2 -Wall -flax-vector-conversions" ./configure  --with-curl
 make -j $(nproc)
-strip -s cpuminer
 mv cpuminer cpuminer-armv8.5-crypto-sha3-sve2
 
 # SHA3 available in armv8.4
 make clean || echo clean
 CFLAGS="-O3 -march=armv8.4-a+crypto+sha3 -Wall -flax-vector-conversions" ./configure  --with-curl
 make -j $(nproc)
-strip -s cpuminer
 mv cpuminer cpuminer-armv8.4-crypto-sha3
 
 make clean || echo clean
 rm -f config.status
 CFLAGS="-O3 -march=armv8-a+crypto -Wall -flax-vector-conversions" ./configure  --with-curl
 make -j $(nproc)
-strip -s cpuminer
 mv cpuminer cpuminer-armv8-crypto
 
 make clean || echo clean
 rm -f config.status
 CFLAGS="-O3 -march=armv8-a -Wall -flax-vector-conversions" ./configure  --with-curl
 make -j $(nproc)
-strip -s cpuminer
 mv cpuminer cpuminer-armv8
 
 make clean || echo clean
 rm -f config.status
 CFLAGS="-O3 -march=native -Wall -flax-vector-conversions" ./configure  --with-curl     
 make -j $(nproc)
-strip -s cpuminer

build-msys2.sh

@@ -1,10 +0,0 @@
-#!/bin/bash
-#
-# Compile on Windows using MSYS2 and MinGW.
-
-make distclean || echo clean
-rm -f config.status
-./autogen.sh || echo done
-CFLAGS="-O3 -march=native -Wall -D_WIN32_WINNT=0x0601" ./configure --with-curl
-make -j $(nproc)
-strip -s cpuminer

build.sh

@@ -1,20 +1,9 @@
-#!/bin/bash
-
-#if [ "$OS" = "Windows_NT" ]; then
-#    ./mingw64.sh
-#    exit 0
-#fi
-
-# Linux build
+#!/bin/sh
 
 make distclean || echo clean
-
 rm -f config.status
 ./autogen.sh || echo done
-
-#CFLAGS="-O3 -march=native -Wall" ./configure --with-curl --with-crypto=$HOME/usr
 CFLAGS="-O3 -march=native -Wall" ./configure --with-curl
-
 make -j $(nproc)
 
-strip -s cpuminer
+#strip -s cpuminer

clean-all.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/bin/sh
 #
 # make clean and rm all the targetted executables.
 

compat.h

@@ -3,7 +3,7 @@
 
 #ifdef WIN32
 
-#if _WIN32_WINNT==0x0601    // Windows 7
+#if _WIN32_WINNT>=0x0601    // Windows 7
  #define WINDOWS_CPU_GROUPS_ENABLED 1
 #endif
 

configure.ac

@@ -1,4 +1,4 @@
-AC_INIT([cpuminer-opt], [24.7])
+AC_INIT([cpuminer-opt], [25.2])
 
 AC_PREREQ([2.59c])
 AC_CANONICAL_SYSTEM
@@ -42,14 +42,11 @@ AC_FUNC_ALLOCA
 AC_CHECK_FUNCS([getopt_long])
 
 case $target in
-  i*86-*-*)
-    have_x86=true
-    ;;
   x86_64-*-*|amd64-*-*)
     have_x86_64=true
     ;;
-  arm*-*-*)
-    have_arm=true
+  aarch64*-*-*|arm64*-*-*)
+    have_arm64=true
     ;;
   powerpc*-*-*)
     have_ppc=true
@@ -65,51 +62,18 @@ case $target in
     PTHREAD_FLAGS=""
     WS2_LIBS="-lws2_32"
     ;;
+  *-apple-*)
+    have_apple=true
+    ;;
 esac
 
-
 AC_ARG_ENABLE([assembly],
   AS_HELP_STRING([--disable-assembly], [disable assembly-language routines]))
 if test x$enable_assembly != xno; then
   AC_DEFINE([USE_ASM], [1], [Define to 1 if assembly routines are wanted.])
 fi
 
-if test x$enable_assembly != xno -a x$have_x86_64 = xtrue
-then
-  AC_MSG_CHECKING(whether we can compile AVX code)
-  AC_COMPILE_IFELSE([AC_LANG_PROGRAM(,[asm ("vmovdqa %ymm0, %ymm1");])],
-    AC_DEFINE(USE_AVX, 1, [Define to 1 if AVX assembly is available.])
-    AC_MSG_RESULT(yes)
-    AC_MSG_CHECKING(whether we can compile XOP code)
-    AC_COMPILE_IFELSE([AC_LANG_PROGRAM(,[asm ("vprotd \$7, %xmm0, %xmm1");])],
-      AC_DEFINE(USE_XOP, 1, [Define to 1 if XOP assembly is available.])
-      AC_MSG_RESULT(yes)
-    ,
-      AC_MSG_RESULT(no)
-      AC_MSG_WARN([The assembler does not support the XOP instruction set.])
-    )
-    AC_MSG_CHECKING(whether we can compile AVX2 code)
-    AC_COMPILE_IFELSE([AC_LANG_PROGRAM(,[asm ("vpaddd %ymm0, %ymm1, %ymm2");])],
-      AC_DEFINE(USE_AVX2, 1, [Define to 1 if AVX2 assembly is available.])
-      AC_MSG_RESULT(yes)
-      AC_MSG_CHECKING(whether we can compile AVX512 code)
-      AC_COMPILE_IFELSE([AC_LANG_PROGRAM(,[asm ("vpaddd %zmm0, %zmm1, %zmm2{%k1}");])],
-        AC_DEFINE(USE_AVX512, 1, [Define to 1 if AVX512 assembly is available.])
-        AC_MSG_RESULT(yes)
-      ,
-        AC_MSG_RESULT(no)
-        AC_MSG_WARN([The assembler does not support the AVX512 instruction set.])
-      )
-    ,
-      AC_MSG_RESULT(no)
-      AC_MSG_WARN([The assembler does not support the AVX2 instruction set.])
-    )
-  ,
-    AC_MSG_RESULT(no)
-    AC_MSG_WARN([The assembler does not support the AVX instruction set.])
-  )
-fi
-
+# jansson test fails on Linux/Mingw, handled in Makefile.am.
 AC_CHECK_LIB(jansson, json_loads, request_jansson=false, request_jansson=true)
 
 AC_CHECK_LIB([pthread], [pthread_create], PTHREAD_LIBS="-lpthread",
@@ -118,9 +82,6 @@ AC_CHECK_LIB([pthread], [pthread_create], PTHREAD_LIBS="-lpthread",
       AC_CHECK_LIB([pthreadGC], [pthread_create], PTHREAD_LIBS="-lpthreadGC"
 ))))
 
-#LDFLAGS="$PTHREAD_LDFLAGS $LDFLAGS"
-# PTHREAD_LIBS="$PTHREAD_LIBS"
-
 AC_MSG_CHECKING(whether __uint128_t is supported)
 AC_COMPILE_IFELSE([AC_LANG_PROGRAM([static __uint128_t i = 100;])],
       AC_DEFINE(USE_INT128, 1, [Define if __uint128_t is available])
@@ -133,16 +94,10 @@ AC_COMPILE_IFELSE([AC_LANG_PROGRAM([static __uint128_t i = 100;])],
 AM_CONDITIONAL([WANT_JANSSON], [test x$request_jansson = xtrue])
 AM_CONDITIONAL([HAVE_WINDOWS], [test x$have_win32 = xtrue])
 AM_CONDITIONAL([USE_ASM], [test x$enable_assembly != xno])
-AM_CONDITIONAL([ARCH_x86], [test x$have_x86 = xtrue])
 AM_CONDITIONAL([ARCH_x86_64], [test x$have_x86_64 = xtrue])
-AM_CONDITIONAL([ARCH_ARM], [test x$have_arm = xtrue])
+AM_CONDITIONAL([ARCH_ARM64], [test x$have_arm64 = xtrue])
 AM_CONDITIONAL([MINGW], [test "x$OS" = "xWindows_NT"])
-
-if test x$request_jansson = xtrue ; then
-	JANSSON_LIBS="compat/jansson/libjansson.a"
-else
-	JANSSON_LIBS=-ljansson
-fi
+AM_CONDITIONAL([HAVE_APPLE], [test x$have_apple = xtrue])
 
 # libcurl install path (for mingw : --with-curl=/usr/local)
 AC_ARG_WITH([curl],
@@ -155,25 +110,10 @@ if test -n "$with_curl" ; then
    LIBCURL="-lcurl -lz"
 fi
 
-# SSL install path (for mingw : --with-crypto=/usr/local/ssl)
-AC_ARG_WITH([crypto],
-   [  --with-crypto=PATH       prefix where openssl crypto is installed [default=/usr]])
-
-if test -n "$with_crypto" ; then
-   LIBCURL_CFLAGS="$LIBCURL_CFLAGS -I$with_crypto/include"
-   LIBCURL_CPPFLAGS="$LIBCURL_CPPFLAGS -I$with_crypto/include"
-   LIBCURL_LDFLAGS="-L$with_crypto/lib $LIBCURL_LDFLAGS"
-   LIBCURL="$LIBCURL -lssl -lcrypto"
-fi
-
 CFLAGS="$CFLAGS $LIBCURL_CFLAGS"
 CPPFLAGS="$CPPFLAGS $LIBCURL_CPPFLAGS"
 LDFLAGS="$LDFLAGS $LIBCURL_LDFLAGS"
 
-#AC_CHECK_LIB([z],[gzopen],[],[])
-#AC_CHECK_LIB([crypto],[OPENSSL_init], crypto=yes, [AC_MSG_ERROR([OpenSSL crypto library required])])
-#AC_CHECK_LIB([ssl],[SSL_new], ssl=yes, ssl=no)
-
 # AC_CHECK_LIB([curl], [curl_multi_timeout],
 #    have_libcurl=yes,
 #    have_libcurl=no AC_MSG_ERROR([curl library required])

configure~

@@ -1,6 +1,6 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.71 for cpuminer-opt 24.7.
+# Generated by GNU Autoconf 2.71 for cpuminer-opt 25.2.
 #
 #
 # Copyright (C) 1992-1996, 1998-2017, 2020-2021 Free Software Foundation,
@@ -608,8 +608,8 @@ MAKEFLAGS=
 # Identity of this package.
 PACKAGE_NAME='cpuminer-opt'
 PACKAGE_TARNAME='cpuminer-opt'
-PACKAGE_VERSION='24.7'
-PACKAGE_STRING='cpuminer-opt 24.7'
+PACKAGE_VERSION='25.2'
+PACKAGE_STRING='cpuminer-opt 25.2'
 PACKAGE_BUGREPORT=''
 PACKAGE_URL=''
 
@@ -657,14 +657,14 @@ JANSSON_LIBS
 LIBCURL_CPPFLAGS
 LIBCURL_CFLAGS
 LIBCURL
+HAVE_APPLE_FALSE
+HAVE_APPLE_TRUE
 MINGW_FALSE
 MINGW_TRUE
-ARCH_ARM_FALSE
-ARCH_ARM_TRUE
+ARCH_ARM64_FALSE
+ARCH_ARM64_TRUE
 ARCH_x86_64_FALSE
 ARCH_x86_64_TRUE
-ARCH_x86_FALSE
-ARCH_x86_TRUE
 USE_ASM_FALSE
 USE_ASM_TRUE
 HAVE_WINDOWS_FALSE
@@ -796,7 +796,6 @@ enable_maintainer_mode
 enable_dependency_tracking
 enable_assembly
 with_curl
-with_crypto
 '
       ac_precious_vars='build_alias
 host_alias
@@ -1360,7 +1359,7 @@ 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 24.7 to adapt to many kinds of systems.
+\`configure' configures cpuminer-opt 25.2 to adapt to many kinds of systems.
 
 Usage: $0 [OPTION]... [VAR=VALUE]...
 
@@ -1432,7 +1431,7 @@ fi
 
 if test -n "$ac_init_help"; then
   case $ac_init_help in
-     short | recursive ) echo "Configuration of cpuminer-opt 24.7:";;
+     short | recursive ) echo "Configuration of cpuminer-opt 25.2:";;
    esac
   cat <<\_ACEOF
 
@@ -1455,7 +1454,6 @@ Optional Packages:
   --with-PACKAGE[=ARG]    use PACKAGE [ARG=yes]
   --without-PACKAGE       do not use PACKAGE (same as --with-PACKAGE=no)
   --with-curl=PATH         prefix where curl is installed default=/usr
-  --with-crypto=PATH       prefix where openssl crypto is installed default=/usr
 
 Some influential environment variables:
   CC          C compiler command
@@ -1538,7 +1536,7 @@ fi
 test -n "$ac_init_help" && exit $ac_status
 if $ac_init_version; then
   cat <<\_ACEOF
-cpuminer-opt configure 24.7
+cpuminer-opt configure 25.2
 generated by GNU Autoconf 2.71
 
 Copyright (C) 2021 Free Software Foundation, Inc.
@@ -1985,7 +1983,7 @@ 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 24.7, which was
+It was created by cpuminer-opt $as_me 25.2, which was
 generated by GNU Autoconf 2.71.  Invocation command line was
 
   $ $0$ac_configure_args_raw
@@ -3593,7 +3591,7 @@ fi
 
 # Define the identity of the package.
  PACKAGE='cpuminer-opt'
- VERSION='24.7'
+ VERSION='25.2'
 
 
 printf "%s\n" "#define PACKAGE \"$PACKAGE\"" >>confdefs.h
@@ -5810,11 +5808,11 @@ if test x$ac_prog_cxx_stdcxx = xno
 then :
   { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for $CXX option to enable C++11 features" >&5
 printf %s "checking for $CXX option to enable C++11 features... " >&6; }
-if test ${ac_cv_prog_cxx_11+y}
+if test ${ac_cv_prog_cxx_cxx11+y}
 then :
   printf %s "(cached) " >&6
 else $as_nop
-  ac_cv_prog_cxx_11=no
+  ac_cv_prog_cxx_cxx11=no
 ac_save_CXX=$CXX
 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
@@ -5856,11 +5854,11 @@ if test x$ac_prog_cxx_stdcxx = xno
 then :
   { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for $CXX option to enable C++98 features" >&5
 printf %s "checking for $CXX option to enable C++98 features... " >&6; }
-if test ${ac_cv_prog_cxx_98+y}
+if test ${ac_cv_prog_cxx_cxx98+y}
 then :
   printf %s "(cached) " >&6
 else $as_nop
-  ac_cv_prog_cxx_98=no
+  ac_cv_prog_cxx_cxx98=no
 ac_save_CXX=$CXX
 cat confdefs.h - <<_ACEOF >conftest.$ac_ext
 /* end confdefs.h.  */
@@ -6502,33 +6500,30 @@ then :
 fi
 
 
-#MINGW_TARGET=`$CC -dumpmachine 2>&1`
-#case $MINGW_TARGET in
 case $target in
-arm*-*-*)
-    have_arm=true
-    ;;
-  i*86-*-mingw*)
-    have_x86=true
-    have_win32=true
-    CFLAGS="-Icompat/pthreads $CFLAGS"
-    PTHREAD_LDFLAGS="-Lcompat/pthreads/x86"
-    WS2_LIBS="-lws2_32"
-    ;;
-  x86_64-*-mingw*|amd64-*-mingw*)
-    have_x86_64=true
-    have_win32=true
-    CFLAGS="-Icompat/pthreads $CFLAGS"
-    PTHREAD_LDFLAGS="-Lcompat/pthreads/x64"
-    # SHOULD BE AT END! after -lcrypto #
-    WS2_LIBS="-L/mingw/x86_64-w64-mingw32/lib -lws2_32"
-    ;;
-  i*86-*-*)
-    have_x86=true
-    ;;
   x86_64-*-*|amd64-*-*)
     have_x86_64=true
     ;;
+  aarch64*-*-*|arm64*-*-*)
+    have_arm64=true
+    ;;
+  powerpc*-*-*)
+    have_ppc=true
+    ;;
+esac
+
+PTHREAD_FLAGS="-pthread"
+WS2_LIBS=""
+
+case $target in
+  *-*-mingw*)
+    have_win32=true
+    PTHREAD_FLAGS=""
+    WS2_LIBS="-lws2_32"
+    ;;
+  *-apple-*)
+    have_apple=true
+    ;;
 esac
 
 # Check whether --enable-assembly was given.
@@ -6543,126 +6538,7 @@ printf "%s\n" "#define USE_ASM 1" >>confdefs.h
 
 fi
 
-if test x$enable_assembly != xno -a x$have_x86_64 = xtrue
-then
-  { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking whether we can compile AVX code" >&5
-printf %s "checking whether we can compile AVX code... " >&6; }
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main (void)
-{
-asm ("vmovdqa %ymm0, %ymm1");
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"
-then :
-
-printf "%s\n" "#define USE_AVX 1" >>confdefs.h
-
-    { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-printf "%s\n" "yes" >&6; }
-    { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking whether we can compile XOP code" >&5
-printf %s "checking whether we can compile XOP code... " >&6; }
-    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main (void)
-{
-asm ("vprotd \$7, %xmm0, %xmm1");
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"
-then :
-
-printf "%s\n" "#define USE_XOP 1" >>confdefs.h
-
-      { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-printf "%s\n" "yes" >&6; }
-
-else $as_nop
-  { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: no" >&5
-printf "%s\n" "no" >&6; }
-      { printf "%s\n" "$as_me:${as_lineno-$LINENO}: WARNING: The assembler does not support the XOP instruction set." >&5
-printf "%s\n" "$as_me: WARNING: The assembler does not support the XOP instruction set." >&2;}
-
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.beam conftest.$ac_ext
-    { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking whether we can compile AVX2 code" >&5
-printf %s "checking whether we can compile AVX2 code... " >&6; }
-    cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main (void)
-{
-asm ("vpaddd %ymm0, %ymm1, %ymm2");
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"
-then :
-
-printf "%s\n" "#define USE_AVX2 1" >>confdefs.h
-
-      { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-printf "%s\n" "yes" >&6; }
-      { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking whether we can compile AVX512 code" >&5
-printf %s "checking whether we can compile AVX512 code... " >&6; }
-      cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-int
-main (void)
-{
-asm ("vpaddd %zmm0, %zmm1, %zmm2{%k1}");
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_compile "$LINENO"
-then :
-
-printf "%s\n" "#define USE_AVX512 1" >>confdefs.h
-
-        { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: yes" >&5
-printf "%s\n" "yes" >&6; }
-
-else $as_nop
-  { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: no" >&5
-printf "%s\n" "no" >&6; }
-        { printf "%s\n" "$as_me:${as_lineno-$LINENO}: WARNING: The assembler does not support the AVX512 instruction set." >&5
-printf "%s\n" "$as_me: WARNING: The assembler does not support the AVX512 instruction set." >&2;}
-
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.beam conftest.$ac_ext
-
-else $as_nop
-  { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: no" >&5
-printf "%s\n" "no" >&6; }
-      { printf "%s\n" "$as_me:${as_lineno-$LINENO}: WARNING: The assembler does not support the AVX2 instruction set." >&5
-printf "%s\n" "$as_me: WARNING: The assembler does not support the AVX2 instruction set." >&2;}
-
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.beam conftest.$ac_ext
-
-else $as_nop
-  { printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: no" >&5
-printf "%s\n" "no" >&6; }
-    { printf "%s\n" "$as_me:${as_lineno-$LINENO}: WARNING: The assembler does not support the AVX instruction set." >&5
-printf "%s\n" "$as_me: WARNING: The assembler does not support the AVX instruction set." >&2;}
-
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
+# jansson test fails on Linux/Mingw, handled in Makefile.am.
 { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for json_loads in -ljansson" >&5
 printf %s "checking for json_loads in -ljansson... " >&6; }
 if test ${ac_cv_lib_jansson_json_loads+y}
@@ -6706,51 +6582,7 @@ else $as_nop
 fi
 
 
-# GC2 for GNU static
-if test "x$have_win32" = "xtrue" ; then
-   # MinGW
-   { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for pthread_create in -lpthread" >&5
-printf %s "checking for pthread_create in -lpthread... " >&6; }
-if test ${ac_cv_lib_pthread_pthread_create+y}
-then :
-  printf %s "(cached) " >&6
-else $as_nop
-  ac_check_lib_save_LIBS=$LIBS
-LIBS="-lpthread  $LIBS"
-cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-
-/* Override any GCC internal prototype to avoid an error.
-   Use char because int might match the return type of a GCC
-   builtin and then its argument prototype would still apply.  */
-char pthread_create ();
-int
-main (void)
-{
-return pthread_create ();
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_link "$LINENO"
-then :
-  ac_cv_lib_pthread_pthread_create=yes
-else $as_nop
-  ac_cv_lib_pthread_pthread_create=no
-fi
-rm -f core conftest.err conftest.$ac_objext conftest.beam \
-    conftest$ac_exeext conftest.$ac_ext
-LIBS=$ac_check_lib_save_LIBS
-fi
-{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_pthread_pthread_create" >&5
-printf "%s\n" "$ac_cv_lib_pthread_pthread_create" >&6; }
-if test "x$ac_cv_lib_pthread_pthread_create" = xyes
-then :
-  PTHREAD_LIBS="-lpthreadGC2"
-fi
-
-else
-   { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for pthread_create in -lpthread" >&5
+{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for pthread_create in -lpthread" >&5
 printf %s "checking for pthread_create in -lpthread... " >&6; }
 if test ${ac_cv_lib_pthread_pthread_create+y}
 then :
@@ -6788,12 +6620,132 @@ printf "%s\n" "$ac_cv_lib_pthread_pthread_create" >&6; }
 if test "x$ac_cv_lib_pthread_pthread_create" = xyes
 then :
   PTHREAD_LIBS="-lpthread"
+else $as_nop
+  { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for pthread_create in -lpthreadGC2" >&5
+printf %s "checking for pthread_create in -lpthreadGC2... " >&6; }
+if test ${ac_cv_lib_pthreadGC2_pthread_create+y}
+then :
+  printf %s "(cached) " >&6
+else $as_nop
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lpthreadGC2  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+char pthread_create ();
+int
+main (void)
+{
+return pthread_create ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"
+then :
+  ac_cv_lib_pthreadGC2_pthread_create=yes
+else $as_nop
+  ac_cv_lib_pthreadGC2_pthread_create=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.beam \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_pthreadGC2_pthread_create" >&5
+printf "%s\n" "$ac_cv_lib_pthreadGC2_pthread_create" >&6; }
+if test "x$ac_cv_lib_pthreadGC2_pthread_create" = xyes
+then :
+  PTHREAD_LIBS="-lpthreadGC2"
+else $as_nop
+  { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for pthread_create in -lpthreadGC1" >&5
+printf %s "checking for pthread_create in -lpthreadGC1... " >&6; }
+if test ${ac_cv_lib_pthreadGC1_pthread_create+y}
+then :
+  printf %s "(cached) " >&6
+else $as_nop
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lpthreadGC1  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+char pthread_create ();
+int
+main (void)
+{
+return pthread_create ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"
+then :
+  ac_cv_lib_pthreadGC1_pthread_create=yes
+else $as_nop
+  ac_cv_lib_pthreadGC1_pthread_create=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.beam \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_pthreadGC1_pthread_create" >&5
+printf "%s\n" "$ac_cv_lib_pthreadGC1_pthread_create" >&6; }
+if test "x$ac_cv_lib_pthreadGC1_pthread_create" = xyes
+then :
+  PTHREAD_LIBS="-lpthreadGC1"
+else $as_nop
+  { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking for pthread_create in -lpthreadGC" >&5
+printf %s "checking for pthread_create in -lpthreadGC... " >&6; }
+if test ${ac_cv_lib_pthreadGC_pthread_create+y}
+then :
+  printf %s "(cached) " >&6
+else $as_nop
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lpthreadGC  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+char pthread_create ();
+int
+main (void)
+{
+return pthread_create ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"
+then :
+  ac_cv_lib_pthreadGC_pthread_create=yes
+else $as_nop
+  ac_cv_lib_pthreadGC_pthread_create=no
+fi
+rm -f core conftest.err conftest.$ac_objext conftest.beam \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ printf "%s\n" "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_pthreadGC_pthread_create" >&5
+printf "%s\n" "$ac_cv_lib_pthreadGC_pthread_create" >&6; }
+if test "x$ac_cv_lib_pthreadGC_pthread_create" = xyes
+then :
+  PTHREAD_LIBS="-lpthreadGC"
+
+fi
+
+fi
+
 fi
 
 fi
 
-LDFLAGS="$PTHREAD_LDFLAGS $LDFLAGS"
-# PTHREAD_LIBS="$PTHREAD_LIBS"
 
 { printf "%s\n" "$as_me:${as_lineno-$LINENO}: checking whether __uint128_t is supported" >&5
 printf %s "checking whether __uint128_t is supported... " >&6; }
@@ -6848,14 +6800,6 @@ else
   USE_ASM_FALSE=
 fi
 
- if test x$have_x86 = xtrue; then
-  ARCH_x86_TRUE=
-  ARCH_x86_FALSE='#'
-else
-  ARCH_x86_TRUE='#'
-  ARCH_x86_FALSE=
-fi
-
  if test x$have_x86_64 = xtrue; then
   ARCH_x86_64_TRUE=
   ARCH_x86_64_FALSE='#'
@@ -6864,12 +6808,12 @@ else
   ARCH_x86_64_FALSE=
 fi
 
- if test x$have_arm = xtrue; then
-  ARCH_ARM_TRUE=
-  ARCH_ARM_FALSE='#'
+ if test x$have_arm64 = xtrue; then
+  ARCH_ARM64_TRUE=
+  ARCH_ARM64_FALSE='#'
 else
-  ARCH_ARM_TRUE='#'
-  ARCH_ARM_FALSE=
+  ARCH_ARM64_TRUE='#'
+  ARCH_ARM64_FALSE=
 fi
 
  if test "x$OS" = "xWindows_NT"; then
@@ -6880,13 +6824,15 @@ else
   MINGW_FALSE=
 fi
 
-
-if test x$request_jansson = xtrue ; then
-	JANSSON_LIBS="compat/jansson/libjansson.a"
+ if test x$have_apple = xtrue; then
+  HAVE_APPLE_TRUE=
+  HAVE_APPLE_FALSE='#'
 else
-	JANSSON_LIBS=-ljansson
+  HAVE_APPLE_TRUE='#'
+  HAVE_APPLE_FALSE=
 fi
 
+
 # libcurl install path (for mingw : --with-curl=/usr/local)
 
 # Check whether --with-curl was given.
@@ -6903,30 +6849,10 @@ if test -n "$with_curl" ; then
    LIBCURL="-lcurl -lz"
 fi
 
-# SSL install path (for mingw : --with-crypto=/usr/local/ssl)
-
-# Check whether --with-crypto was given.
-if test ${with_crypto+y}
-then :
-  withval=$with_crypto;
-fi
-
-
-if test -n "$with_crypto" ; then
-   LIBCURL_CFLAGS="$LIBCURL_CFLAGS -I$with_crypto/include"
-   LIBCURL_CPPFLAGS="$LIBCURL_CPPFLAGS -I$with_crypto/include"
-   LIBCURL_LDFLAGS="-L$with_crypto/lib $LIBCURL_LDFLAGS"
-   LIBCURL="$LIBCURL -lssl -lcrypto"
-fi
-
 CFLAGS="$CFLAGS $LIBCURL_CFLAGS"
 CPPFLAGS="$CPPFLAGS $LIBCURL_CPPFLAGS"
 LDFLAGS="$LDFLAGS $LIBCURL_LDFLAGS"
 
-#AC_CHECK_LIB([z],[gzopen],[],[])
-#AC_CHECK_LIB([crypto],[OPENSSL_init], crypto=yes, [AC_MSG_ERROR([OpenSSL crypto library required])])
-#AC_CHECK_LIB([ssl],[SSL_new], ssl=yes, ssl=no)
-
 # AC_CHECK_LIB([curl], [curl_multi_timeout],
 #    have_libcurl=yes,
 #    have_libcurl=no AC_MSG_ERROR([curl library required])
@@ -7103,22 +7029,22 @@ if test -z "${USE_ASM_TRUE}" && test -z "${USE_ASM_FALSE}"; then
   as_fn_error $? "conditional \"USE_ASM\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
-if test -z "${ARCH_x86_TRUE}" && test -z "${ARCH_x86_FALSE}"; then
-  as_fn_error $? "conditional \"ARCH_x86\" was never defined.
-Usually this means the macro was only invoked conditionally." "$LINENO" 5
-fi
 if test -z "${ARCH_x86_64_TRUE}" && test -z "${ARCH_x86_64_FALSE}"; then
   as_fn_error $? "conditional \"ARCH_x86_64\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
-if test -z "${ARCH_ARM_TRUE}" && test -z "${ARCH_ARM_FALSE}"; then
-  as_fn_error $? "conditional \"ARCH_ARM\" was never defined.
+if test -z "${ARCH_ARM64_TRUE}" && test -z "${ARCH_ARM64_FALSE}"; then
+  as_fn_error $? "conditional \"ARCH_ARM64\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
 if test -z "${MINGW_TRUE}" && test -z "${MINGW_FALSE}"; then
   as_fn_error $? "conditional \"MINGW\" was never defined.
 Usually this means the macro was only invoked conditionally." "$LINENO" 5
 fi
+if test -z "${HAVE_APPLE_TRUE}" && test -z "${HAVE_APPLE_FALSE}"; then
+  as_fn_error $? "conditional \"HAVE_APPLE\" was never defined.
+Usually this means the macro was only invoked conditionally." "$LINENO" 5
+fi
 
 : "${CONFIG_STATUS=./config.status}"
 ac_write_fail=0
@@ -7509,7 +7435,7 @@ cat >>$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 24.7, which was
+This file was extended by cpuminer-opt $as_me 25.2, which was
 generated by GNU Autoconf 2.71.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
@@ -7577,7 +7503,7 @@ ac_cs_config_escaped=`printf "%s\n" "$ac_cs_config" | sed "s/^ //; s/'/'\\\\\\\\
 cat >>$CONFIG_STATUS <<_ACEOF || ac_write_fail=1
 ac_cs_config='$ac_cs_config_escaped'
 ac_cs_version="\\
-cpuminer-opt config.status 24.7
+cpuminer-opt config.status 25.2
 configured by $0, generated by GNU Autoconf 2.71,
   with options \\"\$ac_cs_config\\"
 

cpu-miner.c

@@ -206,7 +206,7 @@ static uint32_t last_block_height = 0;
 static double   highest_share = 0;   // highest accepted share diff
 static double   lowest_share = 9e99; // lowest accepted share diff
 static double   last_targetdiff = 0.;
-#if !(defined(__WINDOWS__) || defined(_WIN64) || defined(_WIN32))
+#if !(defined(__WINDOWS__) || defined(_WIN64) || defined(_WIN32) || defined(__APPLE__))
 static uint32_t hi_temp = 0;
 static uint32_t prev_temp = 0;
 #endif
@@ -286,15 +286,15 @@ static inline void drop_policy(void) { }
 static void affine_to_cpu( struct thr_info *thr )
 {
    int thread = thr->id;
-   unsigned long last_error;    
-   bool ok;
+   unsigned long last_error = 0;    
+   bool ok = true;
 
 #if defined(WINDOWS_CPU_GROUPS_ENABLED)
    unsigned long group_size = GetActiveProcessorCount( 0 );
    unsigned long group      = thread / group_size;
    unsigned long cpu        = thread_affinity_map[ thread % group_size ];
 
-   GROUP_AFFINITY affinity;
+   GROUP_AFFINITY affinity = {0};
    affinity.Group = group;
    affinity.Mask = 1ULL << cpu;
 
@@ -320,8 +320,7 @@ static void affine_to_cpu( struct thr_info *thr )
    {
       last_error = GetLastError();
       if ( !thread )
-      applog( LOG_WARNING, "Set affinity returned error 0x%x for thread %d",
-                           last_error, thread );
+         applog( LOG_WARNING, "Set affinity returned error 0x%x", last_error );
    }
 }   
 
@@ -992,19 +991,19 @@ void report_summary_log( bool force )
 
   if ( rejected_share_count > 10 )
   {
-     if ( rejected_share_count > ( submitted_share_count * .5 ) )
+     if ( rejected_share_count > ( submitted_share_count / 2 ) )
      {
         applog(LOG_ERR,"Excessive rejected share rate, exiting...");
         exit(1);
      } 
-     else if ( rejected_share_count > ( submitted_share_count * .1 ) )
+     else if ( rejected_share_count > ( submitted_share_count / 10 ) )
        applog(LOG_WARNING,"High rejected share rate, check settings.");
    }
 
    gettimeofday( &now, NULL );
    timeval_subtract( &et, &now, &five_min_start );
 
-#if !(defined(__WINDOWS__) || defined(_WIN64) || defined(_WIN32))
+#if !(defined(__WINDOWS__) || defined(_WIN64) || defined(_WIN32) || defined(__APPLE__))
 
    // Display CPU temperature and clock rate.
    int curr_temp = cpu_temp(0); 
@@ -1013,8 +1012,9 @@ void report_summary_log( bool force )
 
    if ( !opt_quiet || ( curr_temp >= 80 ) )
    {
-      int wait_time = curr_temp >= 90 ? 5 : curr_temp >= 80 ? 30 :
-                                            curr_temp >= 70 ? 60 : 120;
+      int wait_time = curr_temp >= 90 ? 5
+                    : curr_temp >= 80 ? 30
+                    : curr_temp >= 70 ? 60 : 120;
       timeval_subtract( &diff, &now, &cpu_temp_time );
       if ( ( diff.tv_sec > wait_time )
         || ( ( curr_temp > prev_temp ) && ( curr_temp >= 75 ) ) )
@@ -1912,7 +1912,7 @@ static bool wanna_mine(int thr_id)
 {
 	bool state = true;
 
-#if !(defined(__WINDOWS__) || defined(_WIN64) || defined(_WIN32))
+#if !(defined(__WINDOWS__) || defined(_WIN64) || defined(_WIN32) || defined(__APPLE__))
   
 	if (opt_max_temp > 0.0)
    {
@@ -2400,7 +2400,7 @@ static void *miner_thread( void *userdata )
              {
                 scale_hash_for_display( &hashrate,  hr_units );
                 sprintf( hr, "%.2f", hashrate );
-#if (defined(_WIN64) || defined(__WINDOWS__) || defined(_WIN32))
+#if (defined(_WIN64) || defined(__WINDOWS__) || defined(_WIN32) || defined(__APPLE__))
                 applog( LOG_NOTICE, "Total: %s %sH/s", hr, hr_units );
 #else
                 float lo_freq = 0., hi_freq = 0.;
@@ -2912,6 +2912,7 @@ static bool cpu_capability( bool display_only )
            sw_arm_arch = __ARM_ARCH;
          #endif
      #endif
+
      // x86_64 only
      #if defined(__SSE2__)
          sw_has_sse2 = true;
@@ -2940,9 +2941,10 @@ static bool cpu_capability( bool display_only )
      #if defined(__AVX10_1_512__)
          sw_has_avx10_512 = true;
      #endif
+
      // x86_64 or AArch64 
      #if defined(__AES__) || defined(__ARM_FEATURE_AES)
-       sw_has_aes = true;
+         sw_has_aes = true;
      #endif
      #ifdef __VAES__
          sw_has_vaes = true;
@@ -2953,6 +2955,7 @@ static bool cpu_capability( bool display_only )
      #if defined(__SHA512__) || defined(__ARM_FEATURE_SHA512)
          sw_has_sha512 = true;
      #endif
+
      // AArch64 only
      #if defined(__ARM_NEON)
          sw_has_neon = true;
@@ -2970,27 +2973,33 @@ static bool cpu_capability( bool display_only )
          sw_has_sme2 = true;
      #endif
 
+     // CPU
      cpu_brand_string( cpu_brand );
      printf( "CPU: %s\n", cpu_brand );
 
-     printf("SW built on " __DATE__
-     #ifdef _MSC_VER
-         " with VC++ 2013\n");
+     // Build
+     printf( "SW built on " __DATE__
+     #if defined(__clang__)
+        " with CLANG-%d.%d.%d", __clang_major__, __clang_minor__,
+                                __clang_patchlevel__ );
      #elif defined(__GNUC__)
-         " with GCC-");
-        printf("%d.%d.%d", __GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__);
-     #else
-        printf("\n");
+        " with GCC-%d.%d.%d", __GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__ );
      #endif
 
+     // OS
      #if defined(__linux)
         printf(" Linux\n");
      #elif defined(WIN32)
-        printf(" Windows\n");
+        printf(" Windows");
+        #if defined(__MINGW64__)
+          printf(" MinGW-w64\n");
+        #else
+          printf("\n");
+        #endif
      #elif defined(__APPLE__)
         printf(" MacOS\n");
-#elif defined(__unix__) || defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__NetBSD__) 
-        printf(" Unix\n");
+     #elif defined(__bsd__) || defined(__unix__) || defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__NetBSD__) 
+        printf(" BSD/Unix\n");
      #else
         printf("\n");
      #endif
@@ -3050,7 +3059,7 @@ static bool cpu_capability( bool display_only )
 
      if ( !display_only )
      {
-        printf("\nAlgo features:       ");
+        printf("\nAlgo features:");
         if ( algo_features == EMPTY_SET ) printf( " None" );
         else
         {
@@ -3058,7 +3067,7 @@ static bool cpu_capability( bool display_only )
            else if ( algo_has_avx2   )  printf( " AVX2  " );
            else if ( algo_has_sse42  )  printf( " SSE4.2" );
            else if ( algo_has_sse2   )  printf( " SSE2  " );
-           if      ( algo_has_neon   )  printf( " NEON  " );
+           if      ( algo_has_neon   )  printf( " NEON"   );
            if      ( algo_has_vaes   )  printf( " VAES"   );
            else if ( algo_has_aes    )  printf( "  AES"   );
            if      ( algo_has_sha512 )  printf( " SHA512" );
@@ -3103,6 +3112,7 @@ static bool cpu_capability( bool display_only )
      return true;
 }
 
+/*
 void show_version_and_exit(void)
 {
         printf("\n built on " __DATE__
@@ -3159,12 +3169,11 @@ void show_version_and_exit(void)
         printf("\n");
         exit(0);
 }
-
+*/
 void show_usage_and_exit(int status)
 {
 	if (status)
                 fprintf(stderr, "Try `--help' for more information.\n");
-//		fprintf(stderr, "Try `" PACKAGE_NAME " --help' for more information.\n");
 	else
 		printf(usage);
 	exit(status);
@@ -3180,7 +3189,6 @@ void parse_arg(int key, char *arg )
 {
 	char *p;
 	int v, i;
-//	uint64_t ul;
 	double d;
 
 	switch( key )
@@ -3324,7 +3332,8 @@ void parse_arg(int key, char *arg )
 		free(rpc_user);
 		rpc_user = strdup(arg);
 		break;
-	case 'o':  // url
+
+   case 'o':  // url
    {
 		char *ap, *hp;
 		ap = strstr( arg, "://" );
@@ -3389,7 +3398,8 @@ void parse_arg(int key, char *arg )
 		have_stratum = !opt_benchmark && !strncasecmp( rpc_url, "stratum", 7 );
 		break;
 	}
-	case 'O':  // userpass
+
+   case 'O':  // userpass
 		p = strchr(arg, ':');
 		if (!p)
       {
@@ -3549,10 +3559,10 @@ void parse_arg(int key, char *arg )
    case 1029:  // stratum-keepalive
       opt_stratum_keepalive = true;
       break;
-   case 'V':
+   case 'V':   // version
       display_cpu_capability();
       exit(0);
-	case 'h':
+	case 'h':   // help
 		show_usage_and_exit(0);
 
    default:
@@ -3691,9 +3701,6 @@ int main(int argc, char *argv[])
 	{
  	   int cpus = GetActiveProcessorCount( i );
 	   num_cpus += cpus;
-
-//	   if (opt_debug)
-//         applog( LOG_INFO, "Found %d CPUs in CPU group %d", cpus, i );
 	}
 
 #else
@@ -3864,12 +3871,23 @@ int main(int argc, char *argv[])
 	}
 #endif
 
-#if defined(WIN32) && defined(WINDOWS_CPU_GROUPS_ENABLED)
-      if ( opt_debug || ( !opt_quiet && num_cpugroups > 1 ) )
-         applog( LOG_INFO, "Found %d CPUs in %d groups",
-                           num_cpus, num_cpugroups );
+#if defined(WIN32)
+
+#if defined(_WIN32_WINNT)
+   if (opt_debug)
+      applog( LOG_INFO, "_WIN232_WINNT = 0x%04x", _WIN32_WINNT ); 
+#else
+   if (opt_debug)
+      applog( LOG_INFO, "_WIN232_WINNT undefined." );
 #endif
-   
+#if defined(WINDOWS_CPU_GROUPS_ENABLED)
+   if ( opt_debug || ( !opt_quiet && num_cpugroups > 1 ) )
+      applog( LOG_INFO, "Found %d CPUs in %d groups",
+                              num_cpus, num_cpugroups );
+#endif
+
+#endif
+
    conditional_state = malloc( opt_n_threads * ((sizeof(bool)) ) );
    memset( conditional_state, 0, opt_n_threads * ((sizeof(bool)) ) );
    
@@ -3890,7 +3908,7 @@ int main(int argc, char *argv[])
          if ( cpu < num_cpus ) active_cpus++;
       }
       if ( opt_n_threads > active_cpus )
-         applog( LOG_WARNING, "Affinity: more threads (%d) than active CPUs (%d)", opt_n_threads, active_cpus );
+         applog( LOG_WARNING, "More miner threads (%d) than active CPUs in affinity mask (%d)", opt_n_threads, active_cpus );
       if ( !opt_quiet )
       {
          char affinity_mask[64];

miner.h

@@ -3,10 +3,7 @@
 
 #include <cpuminer-config.h>
 
-#if !( defined(__SSE2__) || ( defined(__aarch64__) && defined(__ARM_NEON) ) )
-#warning "Unknown or unsupported CPU, requires x86_64 with SSE2 or AArch64 with NEON." 
-#endif
-
+// CPU architecture
 #if defined(__x86_64__)
    #define USER_AGENT_ARCH "x64"     // Intel, AMD x86_64
 #elif defined(__aarch64__)
@@ -17,14 +14,15 @@
    #define USER_AGENT_ARCH
 #endif
 
+// Operating system
+// __APPLE__ includes MacOS & IOS, no MacOS only macros found.
 #if defined(__linux)
    #define USER_AGENT_OS   "L"      // GNU Linux
 #elif defined(WIN32)
    #define USER_AGENT_OS   "W"      // MS Windows
 #elif defined(__APPLE__)
    #define USER_AGENT_OS   "M"      // Apple MacOS
-// is there a generic BSD macro?
-#elif defined(__unix__) || defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__NetBSD__) 
+#elif defined(__bsd__) || defined(__unix__) || defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__NetBSD__) 
    #define USER_AGENT_OS   "U"      // BSD unix
 #else
    #define USER_AGENT_OS

simd-utils.h

@@ -29,7 +29,6 @@
 //    is no significant 64 bit vectorization therefore SSE2 is the practical
 //    minimum for using this code.
 //
-//    MMX:     64 bit vectors  (Not used in cpuminer-opt) 
 //    SSE2:   128 bit vectors  (64 bit CPUs only, such as Intel Core2.
 //    AVX2:   256 bit vectors  (Starting with Intel Haswell and AMD Ryzen)
 //    AVX512: 512 bit vectors  (Starting with SkylakeX)
@@ -217,9 +216,6 @@
 
 #include "simd-utils/simd-int.h"
 
-// x86_64 MMX 64 bit vectors
-#include "simd-utils/simd-64.h"
-
 // x86_64 SSE2 128 bit vectors
 #include "simd-utils/simd-128.h"
 

simd-utils/simd-64.h

@@ -1,193 +0,0 @@
-#if !defined(SIMD_64_H__)
-#define SIMD_64_H__ 1
-
-#if defined(__x86_64__) && defined(__MMX__)
-
-////////////////////////////////////////////////////////////////
-//
-//               64 bit MMX vectors.
-//
-// This code is not used anywhere annd likely never will. It's intent was
-// to support 2 way parallel hashing using  MMX, or NEON for 32 bit hash
-// functions, but hasn't been implementedwas never implemented.
-//
-// MMX is being deprecated by compilers, all intrinsics will be converted to use SSE
-// registers and instructions. MMX will still be available using ASM.
-// For backward compatibility it's likely the compiler won't allow mixing explicit SSE
-// with promoted MMX. It is therefore preferable to implement all 64 bit vector code
-// using explicit SSE with the upper 64 bits being ignored.
-// Using SSE for 64 bit vectors will complicate loading arrays from memory which will
-// always load 128 bits. Odd indexes will need to be extracted from the upper 64 bits
-// of the even index SSE register. 
-// In most cases the exiting 4x32 SSE code can be used with 2 lanes being ignored
-// making ths file obsolete.
-
-
-#define v64_t                        __m64
-#define v64u32_t                     v64_t
-
-#define v64_load                      _mm_load_si64
-#define v64_store                     _mm_store_si64
-
-#define v64_64(i64)                   ((__m64)(i64))
-#define v64_32                        _mm_set1_pi32
-#define v64_16                        _mm_set1_pi16
-#define v64_8                         _mm_set1_pi8
-
-#define v64_add32                     _mm_add_pi32
-#define v64_add16                     _mm_add_pi16
-#define v64_add8                      _mm_add_pi8
-
-#define v64_mul32                     _mm_mullo_pi32
-#define v64_mul16                     _mm_mullo_pi16
-
-// compare
-#define v64_cmpeq32                   _mm_cmpeq_epi32
-#define v64_cmpeq16                   _mm_cmpeq_epi16
-#define v64_cmpeq8                    _mm_cmpeq_epi8
-
-#define v64_cmpgt32                   _mm_cmpgt_epi32
-#define v64_cmpgt16                   _mm_cmpgt_epi16
-#define v64_cmpgt8                    _mm_cmpgt_epi8
-
-#define v64_cmplt32                   _mm_cmplt_epi32
-#define v64_cmplt16                   _mm_cmplt_epi16
-#define v64_cmplt8                    _mm_cmplt_epi8
-
-// bit shift
-#define v64_sl32                      _mm_slli_epi32
-#define v64_sl16                      _mm_slli_epi16
-#define v64_sl8                       _mm_slli_epi8
-
-#define v64_sr32                      _mm_srli_epi32
-#define v64_sr16                      _mm_srli_epi16
-#define v64_sr8                       _mm_srli_epi8
-
-#define v64_sra32                     _mm_srai_epi32
-#define v64_sra16                     _mm_srai_epi16
-#define v64_sra8                      _mm_srai_epi8
-
-#define v64_alignr8                   _mm_alignr_pi8
-#define v64_unpacklo32                _mm_unpacklo_pi32
-#define v64_unpackhi32                _mm_unpackhi_pi32
-#define v64_unpacklo16                _mm_unpacklo_pi16
-#define v64_unpackhi16                _mm_unpacklhi_pi16
-#define v64_unpacklo8                 _mm_unpacklo_pi8
-#define v64_unpackhi8                 _mm_unpackhi_pi16
-
-// Pseudo constants
-
-#define v64_zero        _mm_setzero_si64()
-#define v64_one_64      _mm_set_pi32(  0UL, 1UL )
-#define v64_one_32      v64_32( 1UL )
-#define v64_one_16      v64_16( 1U )
-#define v64_one_8       v64_8(  1U );
-#define v64_neg1        v64_32( 0xFFFFFFFFUL )
-
-#define casti_v64(p,i) (((v64_t*)(p))[(i)])
-
-// Bitwise not: ~(a)
-//#define mm64_not( a ) _mm_xor_si64( (__m64)a, m64_neg1 )
-#define v64_not( a ) ( (v64_t)( ~( (uint64_t)(a) ) )
-
-/*      
-// Unary negate elements
-#define mm64_negate_32( v ) _mm_sub_pi32( m64_zero, v )
-#define mm64_negate_16( v ) _mm_sub_pi16( m64_zero, v )
-#define mm64_negate_8(  v ) _mm_sub_pi8(  m64_zero, v )
-*/
-
-static inline void v64_memset_zero( __m64 *dst,  const int n )
-{   for ( int i = 0; i < n; i++ ) dst[i] = v64_zero; }
-
-static inline void v64_memset( __m64 *dst, const __m64 a, const int n )
-{   for ( int i = 0; i < n; i++ ) dst[i] = a; }
-
-static inline void v64_memcpy( __m64 *dst, const __m64 *src, const int n )
-{   for ( int i = 0; i < n; i ++ ) dst[i] = src[i]; }
-      
-#define v64_or                       _mm_or_si64
-#define v64_and                      _mm_and_si64
-#define v64_xor                      _mm_xor_si64
-#define v64_andnot                   _mm_andnot_si64
-#define v64_xor3( v2, v1, v0 )       v64_xor( v2, v64_andnot( v1, v0 ) )
-#define v64_xorandnot( v2, v1, v0 )  v64_xor( v2, v64_andnot( v1, v0 ) )
-
-
-// Rotate bits in packed elements of 64 bit vector
-#define v64_rol64( a, n ) \
-   _mm_or_si64( _mm_slli_si64( a, n ), \
-                _mm_srli_si64( a, 64-(n) ) )
-
-#define v64_ror64( a, n ) \
-   _mm_or_si64( _mm_srli_si64( a, n ), \
-                _mm_slli_si64( a, 64-(n) ) )
-
-#define v64_rol32( a, n ) \
-   _mm_or_si64( _mm_slli_pi32( a, n ), \
-                _mm_srli_pi32( a, 32-(n) ) )
-
-#define v64_ror32( a, n ) \
-   _mm_or_si64( _mm_srli_pi32( a, n ), \
-                _mm_slli_pi32( a, 32-(n) ) )
-
-#define v64_rol16( a, n ) \
-   _mm_or_si64( _mm_slli_pi16( a, n ), \
-                _mm_srli_pi16( a, 16-(n) ) )
-
-#define v64_ror16( a, n ) \
-   _mm_or_si64( _mm_srli_pi16( a, n ), \
-                _mm_slli_pi16( a, 16-(n) ) )
-
-// Rotate packed elements accross lanes. Useful for byte swap and byte
-// rotation.
-
-#if defined(__SSE__)
-
-// Swap hi & lo 32 bits.
-#define v64_swap32( a )      _mm_shuffle_pi16( a, 0x4e )
-
-#define v64_shulfr16( a )     _mm_shuffle_pi16( a, 0x39 ) 
-#define v64_shufll16( a )     _mm_shuffle_pi16( a, 0x93 ) 
-
-// Swap hi & lo 16 bits of each 32 bit element
-#define v64_swap32_16( a )    _mm_shuffle_pi16( a, 0xb1 )
-
-#endif   // SSE
-
-#if defined(__SSSE3__)
-
-// Endian byte swap packed elements
-
-#define v64_bswap32( v ) \
-    _mm_shuffle_pi8( v, (__m64)0x0405060700010203 )
-
-#define v64_bswap16( v ) \
-    _mm_shuffle_pi8( v, (__m64)0x0607040502030001 );
-
-// Rotate right by c bytes
-static inline v64_t v64_shuflr_x8( __m64 v, const int c )
-{ return _mm_alignr_pi8( v, v, c ); }
-
-#else
-
-#define v64_bswap32( v ) \
-   _mm_set_pi32( __builtin_bswap32( ((uint32_t*)&v)[1] ), \
-                 __builtin_bswap32( ((uint32_t*)&v)[0] )  )
-
-#define v64_bswap16( v ) \
-   _mm_set_pi16( __builtin_bswap16( ((uint16_t*)&v)[3] ), \
-                 __builtin_bswap16( ((uint16_t*)&v)[2] ), \
-                 __builtin_bswap16( ((uint16_t*)&v)[1] ), \
-                 __builtin_bswap16( ((uint16_t*)&v)[0] )  )
-
-#endif   // SSSE3
-
-#define v64_blendv( v1, v0, mask ) \
-   v64_or( v64_and( mask, v1 ), v64_andnot( mask, v0 ) )
-
-
-#endif // MMX
-
-#endif // SIMD_64_H__
-

simd-utils/simd-int.h

@@ -19,6 +19,9 @@ static inline uint64_t bswap_64( uint64_t a )
    return b;
 }
 
+// This produces warnings from clang, but its suggested workaround 
+// "rev32 %w0, %w1\n\t" produced errors instead. GCC doesn't complain and
+// it works as is on both.
 static inline uint32_t bswap_32( uint32_t a )
 {
    uint32_t b;

sysinfos.c

@@ -16,8 +16,8 @@
 #include "miner.h"
 #include "simd-utils.h"
 
-// missing on mingw on arm
-#if defined(__aarch64__) && !defined(WIN32)
+// Missing on MinGW, MacOS
+#if defined(__aarch64__) && !defined(WIN32) && !defined(__APPLE__)
 #define ARM_AUXV 
 #endif
 
@@ -28,7 +28,7 @@
 #include <sys/prctl.h>
 #endif
 
-#ifndef WIN32
+#if !(defined(WIN32) || defined(__APPLE__))
 
 // 1035g1: /sys/devices/platform/coretemp.0/hwmon/hwmon3/temp1_input
 // 1035g1: /sys/class/hwmon/hwmon1/temp1_input wrong temp
@@ -152,7 +152,7 @@ static inline void linux_cpu_hilo_freq( float *lo, float *hi )
 
 static inline float cpu_temp( int core )
 {
-#ifdef WIN32
+#if defined(WIN32) || defined(__APPLE__)
 	return 0.;
 #else
 	return linux_cputemp( core );
@@ -161,7 +161,7 @@ static inline float cpu_temp( int core )
 
 static inline uint32_t cpu_clock( int core )
 {
-#ifdef WIN32
+#if defined(WIN32) || defined(__APPLE__)
 	return 0;
 #else
 	return linux_cpufreq( core );
@@ -280,8 +280,8 @@ static inline int cpu_fanpercent()
 #define FMA3_mask    (FMA3_Flag|AVX_mask)
 #define AVX512_mask  (AVX512_VL_Flag|AVX512_BW_Flag|AVX512_DQ_Flag|AVX512_F_Flag)
 
-
 #if defined(__x86_64__)
+
 static inline void cpuid( unsigned int leaf, unsigned int subleaf,
                           unsigned int output[4] )
 {
@@ -965,7 +965,7 @@ static inline unsigned int avx10_vector_length()
     return 0;
 }
 
-// ARM SVE vector register length
+// ARM SVE vector register length, converted from bytes to bits.
 static inline int sve_vector_length()
 {
 #if defined(ARM_AUXV)

winbuild-cross.sh

@@ -11,16 +11,11 @@
 
 export LOCAL_LIB="$HOME/usr/lib"
 export CONFIGURE_ARGS="--with-curl=$LOCAL_LIB/curl --host=x86_64-w64-mingw32"
-#export CONFIGURE_ARGS="--with-curl=$LOCAL_LIB/curl --with-crypto=$LOCAL_LIB/openssl --host=x86_64-w64-mingw32"
 export MINGW_LIB="/usr/x86_64-w64-mingw32/lib"
 # set correct gcc version
 export GCC_MINGW_LIB="/usr/lib/gcc/x86_64-w64-mingw32/9.3-win32"
 # used by GCC
 export LDFLAGS="-L$LOCAL_LIB/curl/lib/.libs -L$LOCAL_LIB/gmp/.libs"
-#export LDFLAGS="-L$LOCAL_LIB/curl/lib/.libs -L$LOCAL_LIB/gmp/.libs -L$LOCAL_LIB/openssl"
-# Support for Windows 7 CPU groups, AES sometimes not included in -march
-# CPU groups disabled due to incompatibilities between Intel and AMD CPUs.
-#export DEFAULT_CFLAGS="-maes -O3 -Wall -D_WIN32_WINNT=0x0601"
 export DEFAULT_CFLAGS="-maes -O3 -Wall"
 export DEFAULT_CFLAGS_OLD="-O3 -Wall"
 
@@ -40,7 +35,6 @@ cp $MINGW_LIB/zlib1.dll release/
 cp $MINGW_LIB/libwinpthread-1.dll release/
 cp $GCC_MINGW_LIB/libstdc++-6.dll release/
 cp $GCC_MINGW_LIB/libgcc_s_seh-1.dll release/
-#cp ./../libcrypto-1_1-x64.dll release/
 cp $LOCAL_LIB/curl/lib/.libs/libcurl-4.dll release/
 
 # Start building...