v24.7

Makefile.am

@@ -16,6 +16,7 @@ bin_PROGRAMS	= cpuminer
 dist_man_MANS	= cpuminer.1
 
 cpuminer_SOURCES = \
+  dummy.cpp \
   cpu-miner.c \
   util.c \
   api.c \
@@ -113,7 +114,6 @@ cpuminer_SOURCES = \
   algo/lyra2/phi2-4way.c \
   algo/lyra2/phi2.c \
   algo/m7m/m7m.c \
-  algo/m7m/magimath.cpp \
   algo/nist5/nist5-gate.c \
   algo/nist5/nist5-4way.c \
   algo/nist5/nist5.c \
@@ -166,7 +166,6 @@ cpuminer_SOURCES = \
   algo/shavite/sph-shavite-aesni.c \
   algo/shavite/shavite-hash-2way.c \
   algo/shavite/shavite-hash-4way.c \
-  algo/shavite/shavite.c \
   algo/simd/nist.c \
   algo/simd/vector.c \
   algo/simd/sph_simd.c \
@@ -289,14 +288,10 @@ if HAVE_WINDOWS
 endif
 
 cpuminer_LDFLAGS	= @LDFLAGS@
-cpuminer_LDADD	= @LIBCURL@ @JANSSON_LIBS@ @PTHREAD_LIBS@ @WS2_LIBS@  -lgmp
+cpuminer_LDADD	= @LIBCURL@ @JANSSON_LIBS@ @PTHREAD_LIBS@ @WS2_LIBS@ -lgmp
 cpuminer_CPPFLAGS = @LIBCURL_CPPFLAGS@ $(ALL_INCLUDES)
 cpuminer_CFLAGS   = -Wno-pointer-sign -Wno-pointer-to-int-cast $(disable_flags)
 
-if HAVE_WINDOWS
-cpuminer_CFLAGS += -Wl,--stack,10485760
-endif
-
 if HAVE_WINDOWS
 # use to profile an object
 # gprof_cflags = -pg -g3

README.md

@@ -87,7 +87,6 @@ Supported Algorithms
                           groestl       Groestl coin
                           hex           x16r-hex
                           hmq1725       
-                          hodl          Hodlcoin
                           jha           Jackpotcoin
                           keccak        Maxcoin
                           keccakc       Creative coin
@@ -115,9 +114,11 @@ Supported Algorithms
                           scrypt:N      scrypt(N, 1, 1)
                           scryptn2      scrypt(1048576, 1, 1)
                           sha256d       Double SHA-256
+                          sha256dt
                           sha256q       Quad SHA-256
                           sha256t       Triple SHA-256
                           sha3d         Double keccak256 (BSHA3)
+                          sha512256d
                           skein         Skein+Sha (Skeincoin)
                           skein2        Double Skein (Woodcoin)
                           skunk         Signatum (SIGT)
@@ -145,6 +146,7 @@ Supported Algorithms
                           x16rt-veil    veil
                           x16s          
                           x17
+                          x20r
                           x21s
                           x22i
                           x25x

RELEASE_NOTES

@@ -75,6 +75,66 @@ If not what makes it happen or not happen?
 Change Log
 ----------
 
+v24.7
+
+ARM: compile works for Windows using MSys2 & MingW, see wiki for details.
+
+v24.6
+
+ARM: Fixed scryptn2, x16*, broken in v24.2. 
+ARM: Small improvement to interleaving.
+Eliminated some potential compile errors in code that was dependent on 
+compiler optimisations.
+x86_64: improved support for AVX10 compilation, needs GCC-14 or higher.
+
+v24.5
+
+Fix MinGW compile error after MSys2 upgrade to GCC-14.2. 
+#427: GBT: Improved handling of new work.
+Removed shavite3 algo.
+
+v24.4
+
+x86_64: fixed a bug in ornot macro for AVX2 which broke some algos in v24.2.
+x86_64: fixed a bug in alignr macros for SSE2.
+ARM: CPU feature reporting enhancements.
+Some code cleanup.
+
+v24.3
+
+ARM: CPU feature detection and reporting is now working.
+ARM: Verthash is now working.
+ARM: Small speedup for yescrypt, yespower & argon2d.
+Code cleanup.
+
+v24.2
+
+x86_64: Fixed blakes2s for AVX2 & AVX512, x25x for AVX512, broken in v3.23.4.
+x86_64: Initial support for CPUs with AVX10, needs GCC-14.
+ARM NEON: Various code optimisations.
+
+v24.1
+
+#414: fix bug in merkle error handling.
+#416: change $nproc to $(nproc) in build scripts.
+#420: change some inline function definitions to static inline. 
+#413: Fix formatting error for share result log when using no-color.
+Faster 2 way interleaving.
+Cleanup sha256 architecture targetting.
+
+v23.15
+
+Fixed x11gost (sib) algo for all architectures, broken in v3.23.4.
+ARM: Fugue AES optimizations enabled.
+ARM: quark, qubit, x11gost algos optimized with NEON & AES.
+
+v23.14
+
+ARM: Groestl AES optimizations enabled.
+All: Small optimization to Shabal 4way.
+x86_64: Extend Shabal 4way support to SSE2 from SSE4.1.
+All: deleted some unused files.
+
 v23.13
 
 Added x20r algo.

algo-gate-api.c

@@ -184,7 +184,7 @@ int scanhash_4way_64in_32out( struct work *work, uint32_t max_nonce,
 
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 //int scanhash_8way_64_64( struct work *work, uint32_t max_nonce,
 //                      uint64_t *hashes_done, struct thr_info *mythr )
@@ -263,8 +263,8 @@ static void init_algo_gate( algo_gate_t* gate )
    gate->build_block_header      = (void*)&std_build_block_header;
    gate->build_extraheader       = (void*)&std_build_extraheader;
    gate->set_work_data_endian    = (void*)&do_nothing;
-   gate->resync_threads          = (void*)&do_nothing;
-   gate->do_this_thread          = (void*)&return_true;
+//   gate->resync_threads          = (void*)&do_nothing;
+//   gate->do_this_thread          = (void*)&return_true;
    gate->longpoll_rpc_call       = (void*)&std_longpoll_rpc_call;
    gate->get_work_data_size      = (void*)&std_get_work_data_size;
    gate->optimizations           = EMPTY_SET;
@@ -340,7 +340,6 @@ bool register_algo_gate( int algo, algo_gate_t *gate )
     case ALGO_SHA256T:      rc = register_sha256t_algo       ( gate ); break;
     case ALGO_SHA3D:        rc = register_sha3d_algo         ( gate ); break;
     case ALGO_SHA512256D:   rc = register_sha512256d_algo    ( gate ); break;
-    case ALGO_SHAVITE3:     rc = register_shavite_algo       ( gate ); break;
     case ALGO_SKEIN:        rc = register_skein_algo         ( gate ); break;
     case ALGO_SKEIN2:       rc = register_skein2_algo        ( gate ); break;
     case ALGO_SKUNK:        rc = register_skunk_algo         ( gate ); break;

algo-gate-api.h

@@ -98,25 +98,27 @@ typedef  uint32_t set_t;
 #define AVX512_OPT       1 <<  6   // Skylake-X, Zen4 (AVX512[F,VL,DQ,BW])
 #define AES_OPT          1 <<  7   // Intel Westmere, AArch64
 #define VAES_OPT         1 <<  8   // Icelake, Zen3
-#define SHA_OPT          1 <<  9   // Zen1, Icelake, AArch64 
+#define SHA256_OPT       1 <<  9   // Zen1, Icelake, AArch64 
 #define SHA512_OPT       1 << 10   // Intel Arrow Lake, AArch64 
 #define NEON_OPT         1 << 11   // AArch64 
+#define AVX10_256        1 << 12
+#define AVX10_512        1 << 13
 
 // AVX10 does not have explicit algo features:
 //  AVX10_512 is compatible with AVX512 + VAES
 //  AVX10_256 is compatible with AVX2 + VAES
 
 // return set containing all elements from sets a & b
-inline set_t set_union ( set_t a, set_t b ) { return a | b; }
+static inline set_t set_union ( set_t a, set_t b ) { return a | b; }
 
 // return set contained common elements from sets a & b
-inline set_t set_intsec ( set_t a, set_t b) { return a & b; }
+static inline set_t set_intsec ( set_t a, set_t b) { return a & b; }
 
 // all elements in set a are included in set b
-inline bool set_incl ( set_t a, set_t b ) { return (a & b) == a; }
+static inline bool set_incl ( set_t a, set_t b ) { return (a & b) == a; }
 
 // no elements in set a are included in set b
-inline bool set_excl ( set_t a, set_t b ) { return (a & b) == 0; }
+static inline bool set_excl ( set_t a, set_t b ) { return (a & b) == 0; }
 
 typedef struct
 {
@@ -163,10 +165,10 @@ char* ( *malloc_txs_request )   ( struct work* );
 void ( *set_work_data_endian )  ( struct work* );
 
 // Diverge mining threads
-bool ( *do_this_thread )        ( int );
+//bool ( *do_this_thread )        ( int );
 
 // After do_this_thread
-void ( *resync_threads )        ( int, struct work* );
+//void ( *resync_threads )        ( int, struct work* );
 
 json_t* ( *longpoll_rpc_call )  ( CURL*, int*, char* );
 
@@ -246,7 +248,7 @@ int scanhash_4way_64in_32out( struct work *work, uint32_t max_nonce,
 
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 //int scanhash_8way_64in_64out( struct work *work, uint32_t max_nonce,
 //                      uint64_t *hashes_done, struct thr_info *mythr );

algo/argon2d/argon2d/opt.c

@@ -35,7 +35,7 @@
  * @pre all block pointers must be valid
  */
 
-#if defined(__AVX512F__)
+#if defined(SIMD512)
 
 static inline __m512i blamka( __m512i x, __m512i y )
 {
@@ -237,7 +237,7 @@ void fill_segment(const argon2_instance_t *instance,
     uint64_t pseudo_rand, ref_index, ref_lane;
     uint32_t prev_offset, curr_offset;
     uint32_t starting_index, i;
-#if defined(__AVX512F__)
+#if defined(SIMD512)
     __m512i state[ARGON2_512BIT_WORDS_IN_BLOCK];
 #elif defined(__AVX2__)
     __m256i state[ARGON2_HWORDS_IN_BLOCK];

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

@@ -21,7 +21,7 @@
 #include "blake2-impl.h"
 #include "simd-utils.h"
 
-#if !defined(__AVX512F__)
+#if !defined(SIMD512)
 
 #if !defined(__AVX2__)
 

algo/blake/blake-4way.c

@@ -39,7 +39,7 @@ int scanhash_blake_4way( struct work *work, uint32_t max_nonce,
    blake256r14_4way_update( &blake_4w_ctx, vdata, 64 );
 
    do {
-      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
+      *noncev = v128_bswap32( _mm_set_epi32( n+3, n+2, n+1, n ) );
 
       blakehash_4way( hash, vdata );
 

algo/blake/blake256-hash.c

@@ -1611,7 +1611,7 @@ void blake256_8way_final_rounds_le( void *final_hash, const void *midstate,
 
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 ///////////////////////////////////////
 //
@@ -2617,7 +2617,7 @@ blake32_8way_close_le( blake_8way_small_context *sc, unsigned ub, unsigned n,
 #endif
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 //Blake-256 16 way AVX512
 

algo/blake/blake256-hash.h

@@ -147,7 +147,7 @@ void blake256r8_8way_close(void *cc, void *dst);
 #define blake256r8_8x32_update        blake256r14_8way_update
 #define blake256r8_8x32_close         blake256r14_8way_close
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 ///////////////////////////////////
 //

algo/blake/blake2b-hash.c

@@ -226,7 +226,7 @@ static const uint8_t sigma[12][16] =
 #define Mx_(n)      Mx__(n)
 #define Mx__(n)     M ## n
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 #define B2B8W_G(a, b, c, d, x, y) \
 { \
@@ -240,7 +240,7 @@ static const uint8_t sigma[12][16] =
    v[b] = mm512_ror_64( _mm512_xor_si512( v[b], v[c] ), 63 ); \
 }
 
-static void blake2b_8way_compress( blake2b_8way_ctx *ctx, int last )
+static void blake2b_8x64_compress( blake2b_8x64_ctx *ctx, int last )
 {  
    __m512i v[16], m[16];
 
@@ -306,7 +306,7 @@ static void blake2b_8way_compress( blake2b_8way_ctx *ctx, int last )
    ctx->h[7] = mm512_xor3( ctx->h[7], v[7], v[15] );
 }
 
-int blake2b_8way_init( blake2b_8way_ctx *ctx )
+int blake2b_8x64_init( blake2b_8x64_ctx *ctx )
 {
    size_t i;
 
@@ -333,7 +333,7 @@ int blake2b_8way_init( blake2b_8way_ctx *ctx )
 }
 
 
-void blake2b_8way_update( blake2b_8way_ctx *ctx, const void *input,
+void blake2b_8x64_update( blake2b_8x64_ctx *ctx, const void *input,
                           size_t inlen )
 {
    __m512i* in =(__m512i*)input;
@@ -348,7 +348,7 @@ void blake2b_8way_update( blake2b_8way_ctx *ctx, const void *input,
          ctx->t[0] += ctx->c;
          if ( ctx->t[0] < ctx->c )
             ctx->t[1]++;
-         blake2b_8way_compress( ctx, 0 );
+         blake2b_8x64_compress( ctx, 0 );
          ctx->c = 0;
       }
       ctx->b[ c++ ] = in[i];
@@ -356,7 +356,7 @@ void blake2b_8way_update( blake2b_8way_ctx *ctx, const void *input,
    }
 }
 
-void blake2b_8way_final( blake2b_8way_ctx *ctx, void *out )
+void blake2b_8x64_final( blake2b_8x64_ctx *ctx, void *out )
 {
    size_t c;
    c = ctx->c >> 3;
@@ -371,7 +371,7 @@ void blake2b_8way_final( blake2b_8way_ctx *ctx, void *out )
       ctx->c += 8;
    }
 
-   blake2b_8way_compress( ctx, 1 );           // final block flag = 1
+   blake2b_8x64_compress( ctx, 1 );           // final block flag = 1
 
    casti_m512i( out, 0 ) = ctx->h[0];
    casti_m512i( out, 1 ) = ctx->h[1];
@@ -407,7 +407,7 @@ static const uint64_t blake2b_iv[8] = {
 };
 */
 
-static void blake2b_4way_compress( blake2b_4way_ctx *ctx, int last )
+static void blake2b_4x64_compress( blake2b_4x64_ctx *ctx, int last )
 {
 	__m256i v[16], m[16];
 
@@ -473,7 +473,7 @@ static void blake2b_4way_compress( blake2b_4way_ctx *ctx, int last )
    ctx->h[7] = _mm256_xor_si256( _mm256_xor_si256( ctx->h[7], v[7] ), v[15] );
 }
 
-int blake2b_4way_init( blake2b_4way_ctx *ctx ) 
+int blake2b_4x64_init( blake2b_4x64_ctx *ctx ) 
 {
 	size_t i;
 
@@ -499,7 +499,7 @@ int blake2b_4way_init( blake2b_4way_ctx *ctx )
 	return 0;
 }
 
-void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
+void blake2b_4x64_update( blake2b_4x64_ctx *ctx, const void *input,
                           size_t inlen ) 
 {
    __m256i* in =(__m256i*)input;
@@ -514,7 +514,7 @@ void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
 			ctx->t[0] += ctx->c;
 			if ( ctx->t[0] < ctx->c )
 				ctx->t[1]++;
-			blake2b_4way_compress( ctx, 0 );
+			blake2b_4x64_compress( ctx, 0 );
 			ctx->c = 0;
 		}
       ctx->b[ c++ ] = in[i];
@@ -522,7 +522,7 @@ void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
    }
 }
 
-void blake2b_4way_final( blake2b_4way_ctx *ctx, void *out )
+void blake2b_4x64_final( blake2b_4x64_ctx *ctx, void *out )
 {
 	size_t c;
    c = ctx->c >> 3;
@@ -537,7 +537,7 @@ void blake2b_4way_final( blake2b_4way_ctx *ctx, void *out )
       ctx->c += 8;
    }
 
-   blake2b_4way_compress( ctx, 1 );           // final block flag = 1
+   blake2b_4x64_compress( ctx, 1 );           // final block flag = 1
 
    casti_m256i( out, 0 ) = ctx->h[0];
    casti_m256i( out, 1 ) = ctx->h[1];

algo/blake/blake2b-hash.h

@@ -1,6 +1,6 @@
 #pragma once
-#ifndef __BLAKE2B_HASH_4WAY_H__
-#define __BLAKE2B_HASH_4WAY_H__
+#ifndef BLAKE2B_HASH_4WAY_H__
+#define BLAKE2B_HASH_4WAY_H__
 
 #include "simd-utils.h"
 #include <stddef.h>
@@ -15,7 +15,7 @@
 #endif
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct ALIGN( 64 ) {
    __m512i b[16]; // input buffer
@@ -23,12 +23,17 @@ typedef struct ALIGN( 64 ) {
    uint64_t t[2];  // total number of bytes
    size_t c;       // pointer for b[]
    size_t outlen;  // digest size
-} blake2b_8way_ctx;
+} blake2b_8x64_ctx;
 
-int blake2b_8way_init( blake2b_8way_ctx *ctx );
-void blake2b_8way_update( blake2b_8way_ctx *ctx, const void *input,
+int blake2b_8x64_init( blake2b_8x64_ctx *ctx );
+void blake2b_8x64_update( blake2b_8x64_ctx *ctx, const void *input,
                           size_t inlen );
-void blake2b_8way_final( blake2b_8way_ctx *ctx, void *out );
+void blake2b_8x64_final( blake2b_8x64_ctx *ctx, void *out );
+
+#define blake2b_8way_ctx         blake2b_8x64_ctx
+#define blake2b_8way_init        blake2b_8x64_init
+#define blake2b_8way_update      blake2b_8x64_update
+#define blake2b_8way_final       blake2b_8x64_final
 
 #endif
 
@@ -41,12 +46,17 @@ typedef struct ALIGN( 64 ) {
 	uint64_t t[2];  // total number of bytes
 	size_t c;       // pointer for b[]
 	size_t outlen;  // digest size
-} blake2b_4way_ctx;
+} blake2b_4x64_ctx;
 
-int blake2b_4way_init( blake2b_4way_ctx *ctx );
-void blake2b_4way_update( blake2b_4way_ctx *ctx, const void *input,
+int blake2b_4x64_init( blake2b_4x64_ctx *ctx );
+void blake2b_4x64_update( blake2b_4x64_ctx *ctx, const void *input,
                           size_t inlen );
-void blake2b_4way_final( blake2b_4way_ctx *ctx, void *out );
+void blake2b_4x64_final( blake2b_4x64_ctx *ctx, void *out );
+
+#define blake2b_4way_ctx         blake2b_4x64_ctx
+#define blake2b_4way_init        blake2b_4x64_init
+#define blake2b_4way_update      blake2b_4x64_update
+#define blake2b_4way_final       blake2b_4x64_final
 
 #endif
 

algo/blake/blake2b.c

@@ -3,7 +3,7 @@
 #include <stdint.h>
 #include "blake2b-hash.h"
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define BLAKE2B_8WAY
 #elif defined(__AVX2__)
   #define BLAKE2B_4WAY

algo/blake/blake2s-hash.c

@@ -497,7 +497,7 @@ int blake2s_8way_full_blocks( blake2s_8way_state *S, void *out,
 
 #endif // __AVX2__
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // Blake2s-256 16 way
 

algo/blake/blake2s-hash.h

@@ -11,8 +11,8 @@
  * this software. If not, see <http://creativecommons.org/publicdomain/zero/1.0/>.
  */
 //#pragma once
-#ifndef __BLAKE2S_HASH_4WAY_H__
-#define __BLAKE2S_HASH_4WAY_H__ 1
+#ifndef BLAKE2S_HASH_4WAY_H__
+#define BLAKE2S_HASH_4WAY_H__ 1
 
 #if defined(__SSE2__) || defined(__ARM_NEON)
 
@@ -29,20 +29,20 @@
 #define ALIGN(x) __attribute__((aligned(x)))
 #endif
 
-   typedef struct __blake2s_nway_param
-   {
-      uint8_t  digest_length; // 1
-      uint8_t  key_length;    // 2
-      uint8_t  fanout;        // 3
-      uint8_t  depth;         // 4
-      uint32_t leaf_length;   // 8
-      uint8_t  node_offset[6];// 14
-      uint8_t  node_depth;    // 15
-      uint8_t  inner_length;  // 16
-      // uint8_t  reserved[0];
-      uint8_t  salt[8]; // 24
-      uint8_t  personal[8];  // 32
-   } blake2s_nway_param;
+typedef struct __blake2s_nway_param
+{
+   uint8_t  digest_length; // 1
+   uint8_t  key_length;    // 2
+   uint8_t  fanout;        // 3
+   uint8_t  depth;         // 4
+   uint32_t leaf_length;   // 8
+   uint8_t  node_offset[6];// 14
+   uint8_t  node_depth;    // 15
+   uint8_t  inner_length;  // 16
+   // uint8_t  reserved[0];
+   uint8_t  salt[8]; // 24
+   uint8_t  personal[8];  // 32
+} blake2s_nway_param;
 
 typedef struct ALIGN( 64 ) __blake2s_4way_state
 {
@@ -67,7 +67,7 @@ int blake2s_4way_full_blocks( blake2s_4way_state *S, void *out,
 typedef struct ALIGN( 64 ) __blake2s_8way_state
 {
    __m256i h[8];
-   uint8_t  buf[ 32 * 8 ];
+   uint8_t  buf[ 64 * 8 ];
    uint32_t t[2];
    uint32_t f[2];
    size_t   buflen;
@@ -83,12 +83,12 @@ int blake2s_8way_full_blocks( blake2s_8way_state *S, void *out,
 
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct ALIGN( 64 ) __blake2s_16way_state
 {
    __m512i h[8];
-   uint8_t  buf[ 32 * 16 ];
+   uint8_t  buf[ 64 * 16 ];
    uint32_t t[2];
    uint32_t f[2];
    size_t   buflen;

algo/blake/blake2s.c

@@ -3,7 +3,7 @@
 #include <string.h>
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define BLAKE2S_16WAY
 #elif defined(__AVX2__)
   #define BLAKE2S_8WAY

algo/blake/blake512-hash.c

@@ -349,16 +349,16 @@ void blake512_transform( uint64_t *H, const uint64_t *buf, const uint64_t T0,
    Va = v128_add64( Va, v128_add64( Vb, \
                             v128_set64( CBx( r, Sd ) ^ Mx( r, Sc ), \
                                         CBx( r, Sb ) ^ Mx( r, Sa ) ) ) ); \
-   Vd = v128_ror64( v128_xor( Vd, Va ), 32 ); \
+   Vd = v128_ror64xor( Vd, Va, 32 ); \
    Vc = v128_add64( Vc, Vd ); \
-   Vb = v128_ror64( v128_xor( Vb, Vc ), 25 ); \
+   Vb = v128_ror64xor( Vb, Vc, 25 ); \
 \
    Va = v128_add64( Va, v128_add64( Vb, \
                             v128_set64( CBx( r, Sc ) ^ Mx( r, Sd ), \
                                         CBx( r, Sa ) ^ Mx( r, Sb ) ) ) ); \
-   Vd = v128_ror64( v128_xor( Vd, Va ), 16 ); \
+   Vd = v128_ror64xor( Vd, Va, 16 ); \
    Vc = v128_add64( Vc, Vd ); \
-   Vb = v128_ror64( v128_xor( Vb, Vc ), 11 ); \
+   Vb = v128_ror64xor( Vb, Vc, 11 ); \
 }
 
 #define BLAKE512_ROUND( R ) \
@@ -559,7 +559,7 @@ void blake512_full( blake512_context *sc, void *dst, const void *data,
 
 #if defined(__AVX2__)
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 ////////////////////////////////////
 //
@@ -1887,13 +1887,13 @@ blake512_4x64_close(void *cc, void *dst)
 #define GB_2X64( m0, m1, c0, c1, a, b, c, d ) \
 { \
    a = v128_add64( v128_add64( v128_xor( v128_64( c1 ), m0 ), b ), a ); \
-   d = v128_ror64( v128_xor( d, a ), 32 ); \
+   d = v128_ror64xor( d, a, 32 ); \
    c = v128_add64( c, d ); \
-   b = v128_ror64( v128_xor( b, c ), 25 ); \
+   b = v128_ror64xor( b, c, 25 ); \
    a = v128_add64( v128_add64( v128_xor( v128_64( c0 ), m1 ), b ), a ); \
-   d = v128_ror64( v128_xor( d, a ), 16 ); \
+   d = v128_ror64xor( d, a, 16 ); \
    c = v128_add64( c, d ); \
-   b = v128_ror64( v128_xor( b, c ), 11 ); \
+   b = v128_ror64xor( b, c, 11 ); \
 }
 
 #define ROUND_B_2X64(r) \
@@ -2054,9 +2054,9 @@ void blake512_2x64_prehash_part1_le( blake_2x64_big_context *sc,
    // G4 skip nonce
    V0 = v128_add64( v128_add64( v128_xor( v128_64( CB9 ), sc->buf[ 8] ), V5 ),
                                           V0 );
-   VF = v128_ror64( v128_xor( VF, V0 ), 32 );
+   VF = v128_ror64xor( VF, V0, 32 );
    VA = v128_add64( VA, VF );
-   V5 = v128_ror64( v128_xor( V5, VA ), 25 );
+   V5 = v128_ror64xor( V5, VA, 25 );
    V0 = v128_add64( V0, V5 );
 
    GB_2X64( sc->buf[10], sc->buf[11], CBA, CBB, V1, V6, VB, VC );
@@ -2137,9 +2137,9 @@ void blake512_2x64_prehash_part2_le( blake_2x64_big_context *sc, void *hash,
 
    // finish round 0, with the nonce now available 
    V0 = v128_add64( V0, v128_xor( v128_64( CB8 ), M9 ) );
-   VF = v128_ror64( v128_xor( VF, V0 ), 16 );
+   VF = v128_ror64xor( VF, V0, 16 );
    VA = v128_add64( VA, VF );
-   V5 = v128_ror64( v128_xor( V5, VA ), 11 );
+   V5 = v128_ror64xor( V5, VA, 11 );
 
    // Round 1
    // G0
@@ -2147,34 +2147,34 @@ void blake512_2x64_prehash_part2_le( blake_2x64_big_context *sc, void *hash,
 
    // G1
    V1 = v128_add64( V1, V5 );
-   VD = v128_ror64( v128_xor( VD, V1 ), 32 );
+   VD = v128_ror64xor( VD, V1, 32 );
    V9 = v128_add64( V9, VD );
-   V5 = v128_ror64( v128_xor( V5, V9 ), 25 );
+   V5 = v128_ror64xor( V5, V9, 25 );
    V1 = v128_add64( V1, v128_add64( v128_xor( v128_64( CBx(1,2) ), Mx(1,3) ),
                                               V5 ) );
-   VD = v128_ror64( v128_xor( VD, V1 ), 16 );
+   VD = v128_ror64xor( VD, V1, 16 );
    V9 = v128_add64( V9, VD );
-   V5 = v128_ror64( v128_xor( V5, V9 ), 11 );
+   V5 = v128_ror64xor( V5, V9, 11 );
 
    // G2
    V2 = v128_add64( V2, v128_xor( v128_64( CBF ), M9 ) );
-   VE = v128_ror64( v128_xor( VE, V2 ), 32 );
+   VE = v128_ror64xor( VE, V2, 32 );
    VA = v128_add64( VA, VE );
-   V6 = v128_ror64( v128_xor( V6, VA ), 25 );
+   V6 = v128_ror64xor( V6, VA, 25 );
    V2 = v128_add64( V2, v128_add64( v128_xor( v128_64( CB9 ), MF ), V6 ) );
-   VE = v128_ror64( v128_xor( VE, V2 ), 16 );
+   VE = v128_ror64xor( VE, V2, 16 );
    VA = v128_add64( VA, VE );
-   V6 = v128_ror64( v128_xor( V6, VA ), 11 );
+   V6 = v128_ror64xor( V6, VA, 11 );
 
    // G3
-   VF = v128_ror64( v128_xor( VF, V3 ), 32 );
+   VF = v128_ror64xor( VF, V3, 32 );
    VB = v128_add64( VB, VF );
-   V7 = v128_ror64( v128_xor( V7, VB ), 25 );
+   V7 = v128_ror64xor( V7, VB, 25 );
    V3 = v128_add64( V3, v128_add64( v128_xor( v128_64( CBx(1, 6) ), Mx(1, 7) ),
                                               V7 ) );
-   VF = v128_ror64( v128_xor( VF, V3 ), 16 );
+   VF = v128_ror64xor( VF, V3, 16 );
    VB = v128_add64( VB, VF );
-   V7 = v128_ror64( v128_xor( V7, VB ), 11 );
+   V7 = v128_ror64xor( V7, VB, 11 );
 
    // G4, G5, G6, G7
    GB_2X64(Mx(1, 8), Mx(1, 9), CBx(1, 8), CBx(1, 9), V0, V5, VA, VF);

algo/blake/blake512-hash.h

@@ -92,7 +92,7 @@ void blake512_4x64_final_le( blake_4x64_big_context *sc, void *hash,
 #define blake512_4way_prehash_le  blake512_4x64_prehash_le
 #define blake512_4way_final_le    blake512_4x64_final_le
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 ////////////////////////////
 //

algo/blake/blakecoin-4way.c

@@ -182,7 +182,7 @@ int scanhash_blakecoin_4way( struct work *work, uint32_t max_nonce,
    blake256r8_4way_update( &blakecoin_4w_ctx, vdata, 64 );
 
    do {
-      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
+      *noncev = v128_bswap32( _mm_set_epi32( n+3, n+2, n+1, n ) );
       pdata[19] = n;
       blakecoin_4way_hash( hash, vdata );
 

algo/blake/blakecoin-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define BLAKECOIN_16WAY
 #elif defined(__AVX2__)
   #define BLAKECOIN_8WAY

algo/blake/sph_blake2b.c

@@ -101,15 +101,15 @@
 { \
    Va = v128_add64( Va, v128_add64( Vb, \
                  v128_set64( m[ sigmaR[ Sc ] ], m[ sigmaR[ Sa ] ] ) ) ); \
-   Vd = v128_ror64( v128_xor( Vd, Va ), 32 ); \
+   Vd = v128_ror64xor( Vd, Va, 32 ); \
    Vc = v128_add64( Vc, Vd ); \
-   Vb = v128_ror64( v128_xor( Vb, Vc ), 24 ); \
+   Vb = v128_ror64xor( Vb, Vc, 24 ); \
 \
    Va = v128_add64( Va, v128_add64( Vb, \
                  v128_set64( m[ sigmaR[ Sd ] ], m[ sigmaR[ Sb ] ] ) ) ); \
-   Vd = v128_ror64( v128_xor( Vd, Va ), 16 ); \
+   Vd = v128_ror64xor( Vd, Va, 16 ); \
    Vc = v128_add64( Vc, Vd ); \
-   Vb = v128_ror64( v128_xor( Vb, Vc ), 63 ); \
+   Vb = v128_ror64xor( Vb, Vc, 63 ); \
 }
 
 #define BLAKE2B_ROUND( R ) \

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

@@ -87,7 +87,7 @@ void bmw256_8way_close( bmw256_8way_context *ctx, void *dst );
 
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // BMW-256 16 way 32
 
@@ -157,7 +157,7 @@ void bmw512_4way_addbits_and_close(
 
 #endif  // __AVX2__
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // BMW-512 64 bit 8 way
 typedef struct

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

@@ -1057,7 +1057,7 @@ void bmw256_8way_close( bmw256_8way_context *ctx, void *dst )
 
 #endif // __AVX2__
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // BMW-256 16 way 32
 

algo/bmw/bmw512-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define BMW512_8WAY 1
 #elif defined(__AVX2__)
   #define BMW512_4WAY 1

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

@@ -950,7 +950,7 @@ bmw512_4way_addbits_and_close(void *cc, unsigned ub, unsigned n, void *dst)
 
 #endif  // __AVX2__
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // BMW-512 8 WAY
 

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

@@ -26,7 +26,7 @@ static const uint64_t IV512[] =
 0xA5A70E75D65C8A2B, 0xBC796576B1C62456, 0xE7989AF11921C8F7, 0xD43E3B447795D246
 };
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // 4 way 128 is handy to avoid reinterleaving in many algos.
 // If reinterleaving is necessary it may be more efficient to use

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

@@ -6,7 +6,7 @@
 
 #if defined(__AVX2__)
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 struct _cube_4way_context
 {

algo/cubehash/cubehash_sse2.c

@@ -13,7 +13,7 @@ static void transform( cubehashParam *sp )
     int r;
     const int rounds = sp->rounds;
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
     register __m512i x0, x1;
 

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

@@ -11,7 +11,7 @@ static const unsigned int mul2ipt[] __attribute__ ((aligned (64))) =
 };
 */
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 #define ECHO_SUBBYTES4(state, j) \
    state[0][j] = _mm512_aesenc_epi128( state[0][j], k1 ); \

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

@@ -5,7 +5,7 @@
 
 #include "simd-utils.h"
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct
 {

algo/fugue/fugue-aesni.c

@@ -15,237 +15,176 @@
  *
  */
 
-#if defined(__AES__)
-
-#include <x86intrin.h>
+#if ( defined(__SSE4_1__) && defined(__AES__) ) || ( defined(__ARM_NEON) && defined(__ARM_FEATURE_AES) )
 
 #include <memory.h>
 #include "fugue-aesni.h"
 
+static const v128u64_t _supermix1a	__attribute__ ((aligned (16))) =
+   { 0x0202010807020100, 0x0a05000f06010c0b };
 
-MYALIGN const unsigned long long _supermix1a[]	= {0x0202010807020100, 0x0a05000f06010c0b};
-MYALIGN const unsigned long long _supermix1b[]	= {0x0b0d080703060504, 0x0e0a090c050e0f0a};
-MYALIGN const unsigned long long _supermix1c[]	= {0x0402060c070d0003, 0x090a060580808080};
-MYALIGN const unsigned long long _supermix1d[]	= {0x808080800f0e0d0c, 0x0f0e0d0c80808080};
-MYALIGN const unsigned long long _supermix2a[]	= {0x07020d0880808080, 0x0b06010c050e0f0a};
-MYALIGN const unsigned long long _supermix4a[]	= {0x000f0a050c0b0601, 0x0302020404030e09};
-MYALIGN const unsigned long long _supermix4b[]	= {0x07020d08080e0d0d, 0x07070908050e0f0a};
-MYALIGN const unsigned long long _supermix4c[]	= {0x0706050403020000, 0x0302000007060504};
-MYALIGN const unsigned long long _supermix7a[]	= {0x010c0b060d080702, 0x0904030e03000104};
-MYALIGN const unsigned long long _supermix7b[]	= {0x8080808080808080, 0x0504070605040f06};
-//MYALIGN const unsigned long long _k_n[] = {0x4E4E4E4E4E4E4E4E, 0x1B1B1B1B0E0E0E0E};
-//MYALIGN const unsigned char _shift_one_mask[]   = {7, 4, 5, 6, 11, 8, 9, 10, 15, 12, 13, 14, 3, 0, 1, 2};
-//MYALIGN const unsigned char _shift_four_mask[]  = {13, 14, 15, 12, 1, 2, 3, 0, 5, 6, 7, 4, 9, 10, 11, 8};
-//MYALIGN const unsigned char _shift_seven_mask[] = {10, 11, 8, 9, 14, 15, 12, 13, 2, 3, 0, 1, 6, 7, 4, 5};
-//MYALIGN const unsigned char _aes_shift_rows[]   = {0, 5, 10, 15, 4, 9, 14, 3, 8, 13, 2, 7, 12, 1, 6, 11};
-MYALIGN const unsigned int _inv_shift_rows[] = {0x070a0d00, 0x0b0e0104, 0x0f020508, 0x0306090c};
-MYALIGN const unsigned int _mul2mask[] = {0x1b1b0000, 0x00000000, 0x00000000, 0x00000000};
-MYALIGN const unsigned int _mul4mask[] = {0x2d361b00, 0x00000000, 0x00000000, 0x00000000};
-MYALIGN const unsigned int _lsbmask2[] = {0x03030303, 0x03030303, 0x03030303, 0x03030303};
+static const v128u64_t _supermix1b	__attribute__ ((aligned (16))) =
+   { 0x0b0d080703060504, 0x0e0a090c050e0f0a };
 
+static const v128u64_t _supermix1c	__attribute__ ((aligned (16))) =
+   { 0x0402060c070d0003, 0x090a060580808080 };
 
-MYALIGN const unsigned int _IV512[] = {		
-	0x00000000, 0x00000000,	0x7ea50788, 0x00000000,
+static const v128u64_t _supermix1d	__attribute__ ((aligned (16))) =
+   { 0x808080800f0e0d0c, 0x0f0e0d0c80808080 };
+
+static const v128u64_t _supermix2a	__attribute__ ((aligned (16))) =
+   { 0x07020d0880808080, 0x0b06010c050e0f0a };
+
+static const v128u64_t _supermix4a	__attribute__ ((aligned (16))) =
+   { 0x000f0a050c0b0601, 0x0302020404030e09 };
+
+static const v128u64_t _supermix4b	__attribute__ ((aligned (16))) =
+   { 0x07020d08080e0d0d, 0x07070908050e0f0a };
+
+static const v128u64_t _supermix4c	__attribute__ ((aligned (16))) =
+   { 0x0706050403020000, 0x0302000007060504 };
+
+static const v128u64_t _supermix7a	__attribute__ ((aligned (16))) =
+   { 0x010c0b060d080702, 0x0904030e03000104 };
+
+static const v128u64_t _supermix7b	__attribute__ ((aligned (16))) =
+   { 0x8080808080808080, 0x0504070605040f06 };
+
+static const v128u64_t _inv_shift_rows __attribute__ ((aligned (16))) =
+   { 0x0b0e0104070a0d00, 0x0306090c0f020508 };
+
+static const v128u64_t _mul2mask __attribute__ ((aligned (16))) =
+   { 0x000000001b1b0000, 0x0000000000000000 };
+
+static const v128u64_t _mul4mask __attribute__ ((aligned (16))) =
+   { 0x000000002d361b00, 0x0000000000000000 };
+
+static const v128u64_t _lsbmask2 __attribute__ ((aligned (16))) =
+   { 0x0303030303030303, 0x0303030303030303 };
+
+static const uint32_t _IV512[] __attribute__ ((aligned (32))) =
+ {	0x00000000, 0x00000000,	0x7ea50788, 0x00000000,
 	0x75af16e6, 0xdbe4d3c5, 0x27b09aac, 0x00000000,
 	0x17f115d9, 0x54cceeb6, 0x0b02e806, 0x00000000,
 	0xd1ef924a, 0xc9e2c6aa, 0x9813b2dd, 0x00000000,
 	0x3858e6ca, 0x3f207f43, 0xe778ea25, 0x00000000,
-	0xd6dd1f95, 0x1dd16eda, 0x67353ee1, 0x00000000};
+	0xd6dd1f95, 0x1dd16eda, 0x67353ee1, 0x00000000
+ };
 
-#if defined(__SSE4_1__)
+#if defined(__ARM_NEON)
 
-#define PACK_S0(s0, s1, t1)\
-   s0 = _mm_castps_si128(_mm_insert_ps(_mm_castsi128_ps(s0), _mm_castsi128_ps(s1), 0x30))
+#define mask_1000(v)         v128_put32( v, 0, 3 )
 
-#define UNPACK_S0(s0, s1, t1)\
-   s1 = _mm_castps_si128(_mm_insert_ps(_mm_castsi128_ps(s1), _mm_castsi128_ps(s0), 0xc0));\
-   s0 = mm128_mask_32( s0, 8 )
+static const v128u32_t MASK_3321 __attribute__ ((aligned (16))) =
+   { 0x07060504, 0x0b0a0908, 0x0f0e0d0c, 0x0f0e0d0c };
 
-#define CMIX(s1, s2, r1, r2, t1, t2)\
-   t1 = s1;\
-   t1 = _mm_castps_si128(_mm_shuffle_ps(_mm_castsi128_ps(t1), _mm_castsi128_ps(s2), _MM_SHUFFLE(3, 0, 2, 1)));\
-   r1 = _mm_xor_si128(r1, t1);\
-   r2 = _mm_xor_si128(r2, t1);
+static const v128u32_t MASK_3033 __attribute__ ((aligned (16))) =
+   { 0x0f0e0d0c, 0x0f0e0d0c, 0x03020100, 0x0f0e0d0c };
 
-#else   // SSE2
+static const v128u32_t MASK_3303 __attribute__ ((aligned (16))) =
+   { 0x0f0e0d0c, 0x03020100, 0x0f0e0d0c, 0x0f0e0d0c };
 
-#define PACK_S0(s0, s1, t1)\
-   t1 = _mm_shuffle_epi32(s1, _MM_SHUFFLE(0, 3, 3, 3));\
-   s0 = _mm_xor_si128(s0, t1);
+static const v128u32_t MASK_0321 __attribute__ ((aligned (16))) =
+   { 0x07060504, 0x0b0a0908, 0x0f0e0d0c, 0x03020100 };
 
-#define UNPACK_S0(s0, s1, t1)\
-   t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 3, 3));\
-   s1 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s1), _mm_castsi128_ps(t1)));\
-   s0 = mm128_mask_32( s0, 8 )
+#define shuffle_3303(v)      vqtbl1q_u8( v, MASK_3303 )
+#define shuffle_0321(v)      vqtbl1q_u8( v, MASK_0321 )
 
-#define CMIX(s1, s2, r1, r2, t1, t2)\
-   t1 = _mm_shuffle_epi32(s1, 0xf9);\
-   t2 = _mm_shuffle_epi32(s2, 0xcf);\
-   t1 = _mm_xor_si128(t1, t2);\
-   r1 = _mm_xor_si128(r1, t1);\
-   r2 = _mm_xor_si128(r2, t1)
+#define CMIX( s1, s2, r1, r2, t1, t2 ) \
+   t1 = vqtbl1q_u8( s1, MASK_3321 ); \
+   t2 = vqtbl1q_u8( s2, MASK_3033 ); \
+   t1 = v128_xor( t1, t2 ); \
+   r1 = v128_xor( r1, t1 ); \
+   r2 = v128_xor( r2, t1 );
+
+#elif defined(__SSE4_1__)
+
+#define mask_1000(v)         v128_mask32( v, 8 )
+
+#define shuffle_3303(v)      _mm_shuffle_epi32( v, 0xf3 )
+#define shuffle_0321(v)      _mm_shuffle_epi32( v, 0x39 )
+
+#define CMIX( s1, s2, r1, r2, t1, t2 ) \
+   t1 = s1; \
+   t1 = v128_shuffle2_32( t1, s2, _MM_SHUFFLE( 3, 0, 2, 1 ) ); \
+   r1 = v128_xor( r1, t1 ); \
+   r2 = v128_xor( r2, t1 );
 
 #endif
 
-#define TIX256(msg, s10, s8, s24, s0, t1, t2, t3)\
-	t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
-	s10 = _mm_xor_si128(s10, t1);\
-	t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
-	s0 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s0), _mm_castsi128_ps(t1)));\
-	t1 = _mm_slli_si128(t1, 8);\
-	s8 = _mm_xor_si128(s8, t1);\
-	t1 = _mm_shuffle_epi32(s24, _MM_SHUFFLE(3, 3, 0, 3));\
-	s0 = _mm_xor_si128(s0, t1)
+#define PACK_S0( s0, s1, t1 ) \
+ s0 = v128_movlane32( s0, 3, s1, 0 )
 
-
-#define TIX384(msg, s16, s8, s27, s30, s0, s4, t1, t2, t3)\
-	t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
-	s16 = _mm_xor_si128(s16, t1);\
-	t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
-	s0 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s0), _mm_castsi128_ps(t1)));\
-	t1 = _mm_slli_si128(t1, 8);\
-	s8 = _mm_xor_si128(s8, t1);\
-	t1 = _mm_shuffle_epi32(s27, _MM_SHUFFLE(3, 3, 0, 3));\
-	s0 = _mm_xor_si128(s0, t1);\
-	t1 = _mm_shuffle_epi32(s30, _MM_SHUFFLE(3, 3, 0, 3));\
-	s4 = _mm_xor_si128(s4, t1)
+#define UNPACK_S0( s0, s1, t1 ) \
+   s1 = v128_movlane32( s1, 0, s0, 3 ); \
+   s0 = mask_1000( s0 )
 
 #define TIX512(msg, s22, s8, s24, s27, s30, s0, s4, s7, t1, t2, t3)\
-	t1 = _mm_shuffle_epi32(s0, _MM_SHUFFLE(3, 3, 0, 3));\
-	s22 = _mm_xor_si128(s22, t1);\
-	t1 = _mm_castps_si128(_mm_load_ss((float*)msg));\
-	s0 = _mm_castps_si128(_mm_move_ss(_mm_castsi128_ps(s0), _mm_castsi128_ps(t1)));\
-	t1 = _mm_slli_si128(t1, 8);\
-	s8 = _mm_xor_si128(s8, t1);\
-	t1 = _mm_shuffle_epi32(s24, _MM_SHUFFLE(3, 3, 0, 3));\
-	s0 = _mm_xor_si128(s0, t1);\
-	t1 = _mm_shuffle_epi32(s27, _MM_SHUFFLE(3, 3, 0, 3));\
-	s4 = _mm_xor_si128(s4, t1);\
-	t1 = _mm_shuffle_epi32(s30, _MM_SHUFFLE(3, 3, 0, 3));\
-	s7 = _mm_xor_si128(s7, t1)
+	t1 = shuffle_3303( s0 ); \
+	s22 = v128_xor(s22, t1);\
+	t1 = v128_put32( v128_zero, *(uint32_t*)msg, 0 ); \
+	s0 = v128_movlane32( s0, 0, t1, 0 ); \
+	t1 = v128_alignr64( t1, v128_zero, 1 ); \
+	s8 = v128_xor(s8, t1);\
+	t1 = shuffle_3303( s24 ); \
+	s0 = v128_xor(s0, t1);\
+	t1 = shuffle_3303( s27 ); \
+	s4 = v128_xor(s4, t1);\
+	t1 = shuffle_3303( s30 ); \
+	s7 = v128_xor(s7, t1)
 
-#define PRESUPERMIX(t0, t1, t2, t3, t4)\
-   t2 = t0;\
-   t3 = _mm_add_epi8(t0, t0);\
-   t4 = _mm_add_epi8(t3, t3);\
-   t1 = _mm_srli_epi16(t0, 6);\
-   t1 = _mm_and_si128(t1, M128(_lsbmask2));\
-   t3 = _mm_xor_si128(t3, _mm_shuffle_epi8(M128(_mul2mask), t1));\
-   t0 = _mm_xor_si128(t4, _mm_shuffle_epi8(M128(_mul4mask), t1))
-
-/*
-#define PRESUPERMIX(x, t1, s1, s2, t2)\
-	s1 = x;\
-	s2 = _mm_add_epi8(x, x);\
-	t2 = _mm_add_epi8(s2, s2);\
-	t1 = _mm_srli_epi16(x, 6);\
-	t1 = _mm_and_si128(t1, M128(_lsbmask2));\
-	s2 = _mm_xor_si128(s2, _mm_shuffle_epi8(M128(_mul2mask), t1));\
-	x  = _mm_xor_si128(t2, _mm_shuffle_epi8(M128(_mul4mask), t1))
-*/
-
-#define SUBSTITUTE(r0, _t2 )\
-	_t2 = _mm_shuffle_epi8(r0, M128(_inv_shift_rows));\
-	_t2 = _mm_aesenclast_si128( _t2, v128_zero )
+#define SUBSTITUTE( r0, _t2 ) \
+	_t2 = v128_shuffle8( r0, _inv_shift_rows ); \
+	_t2 = v128_aesenclast_nokey( _t2 )
 
 #define SUPERMIX(t0, t1, t2, t3, t4)\
    t2 = t0;\
-   t3 = _mm_add_epi8(t0, t0);\
-   t4 = _mm_add_epi8(t3, t3);\
-   t1 = _mm_srli_epi16(t0, 6);\
-   t1 = _mm_and_si128(t1, M128(_lsbmask2));\
-   t0 = _mm_xor_si128(t4, _mm_shuffle_epi8(M128(_mul4mask), t1)); \
-   t4 = _mm_shuffle_epi8(t2, M128(_supermix1b));\
-   t3 = _mm_xor_si128(t3, _mm_shuffle_epi8(M128(_mul2mask), t1));\
-   t1 = _mm_shuffle_epi8(t4, M128(_supermix1c));\
-   t4 = _mm_xor_si128(t4, t1);\
-   t1 = _mm_shuffle_epi8(t4, M128(_supermix1d));\
-   t4 = _mm_xor_si128(t4, t1);\
-   t1 = _mm_shuffle_epi8(t2, M128(_supermix1a));\
-   t2 = v128_xor3(t2, t3, t0 );\
-   t2 = _mm_shuffle_epi8(t2, M128(_supermix7a));\
+   t3 = v128_add8( t0, t0 ); \
+   t4 = v128_add8( t3, t3 ); \
+   t1 = v128_sr16( t0, 6 ); \
+   t1 = v128_and( t1, _lsbmask2 ); \
+   t0 = v128_xor( t4, v128_shuffle8( _mul4mask, t1 ) ); \
+   t4 = v128_shuffle8( t2, _supermix1b ); \
+   t3 = v128_xor( t3, v128_shuffle8( _mul2mask, t1 ) ); \
+   t1 = v128_shuffle8( t4, _supermix1c ); \
+   t4 = v128_xor( t4, t1 ); \
+   t1 = v128_shuffle8( t4, _supermix1d ); \
+   t4 = v128_xor( t4, t1 ); \
+   t1 = v128_shuffle8( t2, _supermix1a ); \
+   t2 = v128_xor3( t2, t3, t0 ); \
+   t2 = v128_shuffle8( t2, _supermix7a ); \
    t4 = v128_xor3( t4, t1, t2 ); \
-   t2 = _mm_shuffle_epi8(t2, M128(_supermix7b));\
-   t3 = _mm_shuffle_epi8(t3, M128(_supermix2a));\
-   t1 = _mm_shuffle_epi8(t0, M128(_supermix4a));\
-   t0 = _mm_shuffle_epi8(t0, M128(_supermix4b));\
+   t2 = v128_shuffle8( t2, _supermix7b ); \
+   t3 = v128_shuffle8( t3, _supermix2a ); \
+   t1 = v128_shuffle8( t0, _supermix4a ); \
+   t0 = v128_shuffle8( t0, _supermix4b ); \
    t4 = v128_xor3( t4, t2, t1 ); \
-   t0 = _mm_xor_si128(t0, t3);\
-   t4 = v128_xor3(t4, t0, _mm_shuffle_epi8(t0, M128(_supermix4c)));
-
-/*
-#define SUPERMIX(t0, t1, t2, t3, t4)\
-	PRESUPERMIX(t0, t1, t2, t3, t4);\
-	POSTSUPERMIX(t0, t1, t2, t3, t4)
-*/
-
-#define POSTSUPERMIX(t0, t1, t2, t3, t4)\
-	t1 = _mm_shuffle_epi8(t2, M128(_supermix1b));\
-	t4 = t1;\
-	t1 = _mm_shuffle_epi8(t1, M128(_supermix1c));\
-	t4 = _mm_xor_si128(t4, t1);\
-	t1 = _mm_shuffle_epi8(t4, M128(_supermix1d));\
-	t4 = _mm_xor_si128(t4, t1);\
-	t1 = _mm_shuffle_epi8(t2, M128(_supermix1a));\
-	t4 = _mm_xor_si128(t4, t1);\
-	t2 = v128_xor3(t2, t3, t0 );\
-	t2 = _mm_shuffle_epi8(t2, M128(_supermix7a));\
-	t4 = _mm_xor_si128(t4, t2);\
-	t2 = _mm_shuffle_epi8(t2, M128(_supermix7b));\
-	t4 = _mm_xor_si128(t4, t2);\
-	t3 = _mm_shuffle_epi8(t3, M128(_supermix2a));\
-	t1 = _mm_shuffle_epi8(t0, M128(_supermix4a));\
-	t4 = _mm_xor_si128(t4, t1);\
-	t0 = _mm_shuffle_epi8(t0, M128(_supermix4b));\
-	t0 = _mm_xor_si128(t0, t3);\
-	t4 = _mm_xor_si128(t4, t0);\
-	t0 = _mm_shuffle_epi8(t0, M128(_supermix4c));\
-	t4 = _mm_xor_si128(t4, t0)
-
-#define SUBROUND512_3(r1a, r1b, r1c, r1d, r2a, r2b, r2c, r2d, r3a, r3b, r3c, r3d)\
-	CMIX(r1a, r1b, r1c, r1d, _t0, _t1);\
-	PACK_S0(r1c, r1a, _t0);\
-	SUBSTITUTE(r1c, _t2 );\
-	SUPERMIX(_t2, _t3, _t0, _t1, r1c);\
-	_t0 = _mm_shuffle_epi32(r1c, 0x39);\
-	r2c = _mm_xor_si128(r2c, _t0);\
-   _t0 = mm128_mask_32( _t0, 8 ); \
-	r2d = _mm_xor_si128(r2d, _t0);\
-	UNPACK_S0(r1c, r1a, _t3);\
-	SUBSTITUTE(r2c, _t2 );\
-	SUPERMIX(_t2, _t3, _t0, _t1, r2c);\
-	_t0 = _mm_shuffle_epi32(r2c, 0x39);\
-	r3c = _mm_xor_si128(r3c, _t0);\
-   _t0 = mm128_mask_32( _t0, 8 ); \
-	r3d = _mm_xor_si128(r3d, _t0);\
-	UNPACK_S0(r2c, r2a, _t3);\
-	SUBSTITUTE(r3c, _t2 );\
-	SUPERMIX(_t2, _t3, _t0, _t1, r3c);\
-	UNPACK_S0(r3c, r3a, _t3)
+   t0 = v128_xor( t0, t3 ); \
+   t4 = v128_xor3( t4, t0, v128_shuffle8( t0, _supermix4c ) );
 
 #define SUBROUND512_4(r1a, r1b, r1c, r1d, r2a, r2b, r2c, r2d, r3a, r3b, r3c, r3d, r4a, r4b, r4c, r4d)\
 	CMIX(r1a, r1b, r1c, r1d, _t0, _t1);\
 	PACK_S0(r1c, r1a, _t0);\
 	SUBSTITUTE( r1c, _t2 );\
 	SUPERMIX(_t2, _t3, _t0, _t1, r1c);\
-	_t0 = _mm_shuffle_epi32(r1c, 0x39);\
-	r2c = _mm_xor_si128(r2c, _t0);\
-   _t0 = mm128_mask_32( _t0, 8 ); \
-	r2d = _mm_xor_si128(r2d, _t0);\
+	_t0 = shuffle_0321( r1c ); \
+	r2c = v128_xor(r2c, _t0);\
+   _t0 = mask_1000( _t0 ); \
+	r2d = v128_xor(r2d, _t0);\
 	UNPACK_S0(r1c, r1a, _t3);\
 	SUBSTITUTE(r2c, _t2 );\
 	SUPERMIX(_t2, _t3, _t0, _t1, r2c);\
-	_t0 = _mm_shuffle_epi32(r2c, 0x39);\
-	r3c = _mm_xor_si128(r3c, _t0);\
-   _t0 = mm128_mask_32( _t0, 8 ); \
-	r3d = _mm_xor_si128(r3d, _t0);\
+	_t0 = shuffle_0321( r2c ); \
+	r3c = v128_xor(r3c, _t0);\
+   _t0 = mask_1000( _t0 ); \
+	r3d = v128_xor(r3d, _t0);\
 	UNPACK_S0(r2c, r2a, _t3);\
 	SUBSTITUTE( r3c, _t2 );\
 	SUPERMIX(_t2, _t3, _t0, _t1, r3c);\
-	_t0 = _mm_shuffle_epi32(r3c, 0x39);\
-	r4c = _mm_xor_si128(r4c, _t0);\
-   _t0 = mm128_mask_32( _t0, 8 ); \
-	r4d = _mm_xor_si128(r4d, _t0);\
+	_t0 = shuffle_0321( r3c ); \
+	r4c = v128_xor(r4c, _t0);\
+   _t0 = mask_1000( _t0 ); \
+	r4d = v128_xor(r4d, _t0);\
 	UNPACK_S0(r3c, r3a, _t3);\
 	SUBSTITUTE( r4c, _t2 );\
 	SUPERMIX(_t2, _t3, _t0, _t1, r4c);\
@@ -256,18 +195,19 @@ MYALIGN const unsigned int _IV512[] = {
 	block[1] = col[(base + a + 1) % s];\
 	block[2] = col[(base + a + 2) % s];\
 	block[3] = col[(base + a + 3) % s];\
-	x = _mm_load_si128((__m128i*)block)
+	x = v128_load( (v128_t*)block )
 
 #define STORECOLUMN(x, s)\
-	_mm_store_si128((__m128i*)block, x);\
+	v128_store((v128_t*)block, x );\
 	col[(base + 0) % s] = block[0];\
 	col[(base + 1) % s] = block[1];\
 	col[(base + 2) % s] = block[2];\
 	col[(base + 3) % s] = block[3]
 
-void Compress512(hashState_fugue *ctx, const unsigned char *pmsg, unsigned int uBlockCount)
+void Compress512( hashState_fugue *ctx, const unsigned char *pmsg,
+                  unsigned int uBlockCount )
 {
-   __m128i _t0, _t1, _t2, _t3;
+   v128_t _t0, _t1, _t2, _t3;
 
    switch(ctx->base)
    {
@@ -346,134 +286,133 @@ void Compress512(hashState_fugue *ctx, const unsigned char *pmsg, unsigned int u
       pmsg += 4;
       uBlockCount--;
    }
-
 }
 
-void Final512(hashState_fugue *ctx, BitSequence *hashval)
+void Final512( hashState_fugue *ctx, uint8_t *hashval )
 {
    unsigned int block[4] __attribute__ ((aligned (32)));
    unsigned int col[36] __attribute__ ((aligned (16)));
 	unsigned int i, base;
-	__m128i r0, _t0, _t1, _t2, _t3;
+	v128_t r0, _t0, _t1, _t2, _t3;
 
-	for(i = 0; i < 12; i++)
+	for( i = 0; i < 12; i++ )
 	{
-		_mm_store_si128((__m128i*)block, ctx->state[i]);
+		v128_store( (v128_t*)block, ctx->state[i] );
 
 		col[3 * i + 0] = block[0];
 		col[3 * i + 1] = block[1];
 		col[3 * i + 2] = block[2];
 	}
 
-	base = (36 - (12 * ctx->base)) % 36;
+	base = ( 36 - (12 * ctx->base) ) % 36;
 
-	for(i = 0; i < 32; i++)
+	for( i = 0; i < 32; i++ )
 	{
 		// ROR3
 		base = (base + 33) % 36;
 
 		// CMIX
-		col[(base +  0) % 36] ^= col[(base + 4) % 36];
-		col[(base +  1) % 36] ^= col[(base + 5) % 36];
-		col[(base +  2) % 36] ^= col[(base + 6) % 36];
-		col[(base +  18) % 36] ^= col[(base + 4) % 36];
-		col[(base +  19) % 36] ^= col[(base + 5) % 36];
-		col[(base +  20) % 36] ^= col[(base + 6) % 36];
+		col[ (base +  0) % 36 ] ^= col[ (base + 4) % 36 ];
+		col[ (base +  1) % 36 ] ^= col[ (base + 5) % 36 ];
+		col[ (base +  2) % 36 ] ^= col[ (base + 6) % 36 ];
+		col[ (base + 18) % 36 ] ^= col[ (base + 4) % 36 ];
+		col[ (base + 19) % 36 ] ^= col[ (base + 5) % 36 ];
+		col[ (base + 20) % 36 ] ^= col[ (base + 6) % 36 ];
 
 		// SMIX
-		LOADCOLUMN(r0, 36, 0);
-		SUBSTITUTE(r0, _t2);
-		SUPERMIX(_t2, _t3, _t0, _t1, r0);
-		STORECOLUMN(r0, 36);
+		LOADCOLUMN( r0, 36, 0 );
+		SUBSTITUTE( r0, _t2 );
+		SUPERMIX( _t2, _t3, _t0, _t1, r0 );
+		STORECOLUMN( r0, 36 );
 	}
 
-	for(i = 0; i < 13; i++)
+	for( i = 0; i < 13; i++ )
 	{
 		// S4 += S0; S9 += S0; S18 += S0; S27 += S0;
-		col[(base +  4) % 36] ^= col[(base + 0) % 36];
-		col[(base +  9) % 36] ^= col[(base + 0) % 36];
-		col[(base + 18) % 36] ^= col[(base + 0) % 36];
-		col[(base + 27) % 36] ^= col[(base + 0) % 36];
+		col[ (base +  4) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base +  9) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 18) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 27) % 36 ] ^= col[ (base + 0) % 36 ];
 
 		// ROR9
 		base = (base + 27) % 36;
 
 		// SMIX
-		LOADCOLUMN(r0, 36, 0);
-		SUBSTITUTE(r0, _t2);
-		SUPERMIX(_t2, _t3, _t0, _t1, r0);
-		STORECOLUMN(r0, 36);
+		LOADCOLUMN( r0, 36, 0 );
+		SUBSTITUTE( r0, _t2 );
+		SUPERMIX( _t2, _t3, _t0, _t1, r0 );
+		STORECOLUMN( r0, 36 );
 
 		// S4 += S0; S10 += S0; S18 += S0; S27 += S0;
-		col[(base +  4) % 36] ^= col[(base + 0) % 36];
-		col[(base + 10) % 36] ^= col[(base + 0) % 36];
-		col[(base + 18) % 36] ^= col[(base + 0) % 36];
-		col[(base + 27) % 36] ^= col[(base + 0) % 36];
+		col[ (base +  4) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 10) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 18) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 27) % 36 ] ^= col[ (base + 0) % 36 ];
 
 		// ROR9
 		base = (base + 27) % 36;
 
 		// SMIX
-		LOADCOLUMN(r0, 36, 0);
-		SUBSTITUTE(r0, _t2);
-		SUPERMIX(_t2, _t3, _t0, _t1, r0);
-		STORECOLUMN(r0, 36);
+		LOADCOLUMN( r0, 36, 0 );
+		SUBSTITUTE( r0, _t2 );
+		SUPERMIX( _t2, _t3, _t0, _t1, r0 );
+		STORECOLUMN( r0, 36 );
 
 		// S4 += S0; S10 += S0; S19 += S0; S27 += S0;
-		col[(base +  4) % 36] ^= col[(base + 0) % 36];
-		col[(base + 10) % 36] ^= col[(base + 0) % 36];
-		col[(base + 19) % 36] ^= col[(base + 0) % 36];
-		col[(base + 27) % 36] ^= col[(base + 0) % 36];
+		col[ (base +  4) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 10) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 19) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 27) % 36 ] ^= col[ (base + 0) % 36 ];
 
 		// ROR9
 		base = (base + 27) % 36;
 
 		// SMIX
-		LOADCOLUMN(r0, 36, 0);
-		SUBSTITUTE(r0, _t2);
-		SUPERMIX(_t2, _t3, _t0, _t1, r0);
-		STORECOLUMN(r0, 36);
+		LOADCOLUMN( r0, 36, 0 );
+		SUBSTITUTE( r0, _t2 );
+		SUPERMIX( _t2, _t3, _t0, _t1, r0 );
+		STORECOLUMN( r0, 36 );
 
 		// S4 += S0; S10 += S0; S19 += S0; S28 += S0;
-		col[(base +  4) % 36] ^= col[(base + 0) % 36];
-		col[(base + 10) % 36] ^= col[(base + 0) % 36];
-		col[(base + 19) % 36] ^= col[(base + 0) % 36];
-		col[(base + 28) % 36] ^= col[(base + 0) % 36];
+		col[ (base +  4) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 10) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 19) % 36 ] ^= col[ (base + 0) % 36 ];
+		col[ (base + 28) % 36 ] ^= col[ (base + 0) % 36 ];
 
 		// ROR8
 		base = (base + 28) % 36;
 
 		// SMIX
-		LOADCOLUMN(r0, 36, 0);
-		SUBSTITUTE(r0, _t2);
-		SUPERMIX(_t2, _t3, _t0, _t1, r0);
-		STORECOLUMN(r0, 36);
+		LOADCOLUMN( r0, 36, 0 );
+		SUBSTITUTE( r0, _t2 );
+		SUPERMIX( _t2, _t3, _t0, _t1, r0 );
+		STORECOLUMN( r0, 36 );
 	}
 
 	// S4 += S0; S9 += S0; S18 += S0; S27 += S0;
-	col[(base +  4) % 36] ^= col[(base + 0) % 36];
-	col[(base +  9) % 36] ^= col[(base + 0) % 36];
-	col[(base + 18) % 36] ^= col[(base + 0) % 36];
-	col[(base + 27) % 36] ^= col[(base + 0) % 36];
+	col[ (base +  4) % 36 ] ^= col[ (base + 0) % 36 ];
+	col[ (base +  9) % 36 ] ^= col[ (base + 0) % 36 ];
+	col[ (base + 18) % 36 ] ^= col[ (base + 0) % 36 ];
+	col[ (base + 27) % 36 ] ^= col[ (base + 0) % 36 ];
 
 	// Transform to the standard basis and store output; S1 || S2 || S3 || S4
-	LOADCOLUMN(r0, 36, 1);
-	_mm_store_si128((__m128i*)hashval, r0);
+	LOADCOLUMN( r0, 36, 1 );
+	v128_store( (v128_t*)hashval, r0 );
 
 	// Transform to the standard basis and store output; S9 || S10 || S11 || S12
-	LOADCOLUMN(r0, 36, 9);
-	_mm_store_si128((__m128i*)hashval + 1, r0);
+	LOADCOLUMN( r0, 36, 9 );
+	v128_store( (v128_t*)hashval + 1, r0 );
 
 	// Transform to the standard basis and store output; S18 || S19 || S20 || S21
-	LOADCOLUMN(r0, 36, 18);
-	_mm_store_si128((__m128i*)hashval + 2, r0);
+	LOADCOLUMN( r0, 36, 18 );
+	v128_store( (v128_t*)hashval + 2, r0 );
 
 	// Transform to the standard basis and store output; S27 || S28 || S29 || S30
-	LOADCOLUMN(r0, 36, 27);
-	_mm_store_si128((__m128i*)hashval + 3, r0);
+	LOADCOLUMN( r0, 36, 27 );
+	v128_store( (v128_t*)hashval + 3, r0 );
 }
 
-HashReturn fugue512_Init(hashState_fugue *ctx, int nHashSize)
+int fugue512_Init( hashState_fugue *ctx, int nHashSize )
 {
 	int i;
 	ctx->processed_bits = 0;
@@ -487,18 +426,18 @@ HashReturn fugue512_Init(hashState_fugue *ctx, int nHashSize)
 	for(i = 0; i < 6; i++)
 		ctx->state[i] = v128_zero;
 
-	ctx->state[6]  = _mm_load_si128((__m128i*)_IV512 + 0);
-	ctx->state[7]  = _mm_load_si128((__m128i*)_IV512 + 1);
-	ctx->state[8]  = _mm_load_si128((__m128i*)_IV512 + 2);
-	ctx->state[9]  = _mm_load_si128((__m128i*)_IV512 + 3);
-	ctx->state[10] = _mm_load_si128((__m128i*)_IV512 + 4);
-	ctx->state[11] = _mm_load_si128((__m128i*)_IV512 + 5);
+	ctx->state[6]  = casti_v128( _IV512, 0 );
+	ctx->state[7]  = casti_v128( _IV512, 1 );
+	ctx->state[8]  = casti_v128( _IV512, 2 );
+	ctx->state[9]  = casti_v128( _IV512, 3 );
+	ctx->state[10] = casti_v128( _IV512, 4 );
+	ctx->state[11] = casti_v128( _IV512, 5 );
 
-	return SUCCESS;
+	return 0;
 }
 
-
-HashReturn fugue512_Update(hashState_fugue *state, const void *data, DataLength databitlen)
+int fugue512_Update( hashState_fugue *state, const void *data,
+                            uint64_t databitlen )
 {
 	unsigned int uByteLength, uBlockCount, uRemainingBytes;
 
@@ -509,7 +448,8 @@ HashReturn fugue512_Update(hashState_fugue *state, const void *data, DataLength
 		if(state->uBufferBytes != 0)
 		{
 			// Fill the buffer
-			memcpy(state->buffer + state->uBufferBytes, (void*)data, state->uBlockLength - state->uBufferBytes);
+			memcpy( state->buffer + state->uBufferBytes, (void*)data,
+                 state->uBlockLength - state->uBufferBytes );
 
 			// Process the buffer
 			Compress512(state, state->buffer, 1);
@@ -545,13 +485,13 @@ HashReturn fugue512_Update(hashState_fugue *state, const void *data, DataLength
 		state->uBufferBytes += uByteLength;
 	}
 
-	return SUCCESS;
+	return 0;
 }
 
-HashReturn fugue512_Final(hashState_fugue *state, void *hashval)
+int fugue512_Final( hashState_fugue *state, void *hashval )
 {
 	unsigned int i;
-	BitSequence lengthbuf[8] __attribute__((aligned(64)));
+	uint8_t lengthbuf[8] __attribute__((aligned(64)));
 
 	// Update message bit count
 	state->processed_bits += state->uBufferBytes * 8;
@@ -575,16 +515,17 @@ HashReturn fugue512_Final(hashState_fugue *state, void *hashval)
 	// Finalization
 	Final512(state, hashval);
 
-	return SUCCESS;
+	return 0;
 }
 
 
-HashReturn fugue512_full(hashState_fugue *hs, void *hashval, const void *data, DataLength databitlen)
+int fugue512_full( hashState_fugue *hs, void *hashval, const void *data,
+                   uint64_t databitlen )
 {
-	fugue512_Init(hs, 512);
-	fugue512_Update(hs, data, databitlen*8);
-	fugue512_Final(hs, hashval);
-	return SUCCESS;
+	fugue512_Init( hs, 512 );
+	fugue512_Update( hs, data, databitlen*8 );
+	fugue512_Final( hs, hashval );
+	return 0;
 }
 
 #endif  // AES

algo/fugue/fugue-aesni.h

@@ -14,37 +14,31 @@
 #ifndef FUGUE_HASH_API_H
 #define FUGUE_HASH_API_H
 
-#if defined(__AES__) 
+#if ( defined(__SSE4_1__) && defined(__AES__) ) || ( defined(__ARM_NEON) && defined(__ARM_FEATURE_AES) )
 
-#if !defined(__SSE4_1__)
-#error "Unsupported configuration, AES needs SSE4.1. Compile without AES."
-#endif
-
-#include "compat/sha3_common.h"
 #include "simd-utils.h"
 
-
 typedef struct
 {
-	__m128i			state[12];
+	v128_t			state[12];
 	unsigned int	base;
-
 	unsigned int	uHashSize;
 	unsigned int	uBlockLength;
 	unsigned int	uBufferBytes;
-	DataLength		processed_bits;
-	BitSequence		buffer[4];
+	uint64_t 		processed_bits;
+	uint8_t  		buffer[4];
 
 } hashState_fugue __attribute__ ((aligned (64)));
 
 
 // These functions are deprecated, use the lower case macro aliases that use
 // the standard interface. This will be cleaned up at a later date.
-HashReturn fugue512_Init(hashState_fugue *state, int hashbitlen);
+int fugue512_Init( hashState_fugue *state, int hashbitlen );
 
-HashReturn fugue512_Update(hashState_fugue *state, const void *data, DataLength databitlen);
+int fugue512_Update( hashState_fugue *state, const void *data,
+                     uint64_t databitlen );
 
-HashReturn fugue512_Final(hashState_fugue *state, void *hashval);
+int fugue512_Final( hashState_fugue *state, void *hashval );
 
 #define fugue512_init( state ) \
    fugue512_Init( state, 512 )
@@ -54,7 +48,8 @@ HashReturn fugue512_Final(hashState_fugue *state, void *hashval);
    fugue512_Final
 
 
-HashReturn fugue512_full(hashState_fugue *hs, void *hashval, const void *data, DataLength databitlen);
+int fugue512_full( hashState_fugue *hs, void *hashval, const void *data,
+                   uint64_t databitlen);
 
 #endif // AES
 #endif // HASH_API_H

algo/gost/sph_gost.c

@@ -696,7 +696,7 @@ static void AddModulo512(const void *a,const void *b,void *c)
 
 static void AddXor512(const void *a,const void *b,void *c)
 {
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
    casti_m512i( c, 0 ) = _mm512_xor_si512( casti_m512i( a, 0 ),
                                            casti_m512i( b, 0 ) );
 #elif defined(__AVX2__)
@@ -704,15 +704,15 @@ static void AddXor512(const void *a,const void *b,void *c)
                                            casti_m256i( b, 0 ) );
    casti_m256i( c, 1 ) = _mm256_xor_si256( casti_m256i( a, 1 ),
                                            casti_m256i( b, 1 ) );
-#elif defined(__SSE2__)
-   casti_m128i( c, 0 ) = _mm_xor_si128( casti_m128i( a, 0 ),
-                                        casti_m128i( b, 0 ) );
-   casti_m128i( c, 1 ) = _mm_xor_si128( casti_m128i( a, 1 ),
-                                        casti_m128i( b, 1 ) );
-   casti_m128i( c, 2 ) = _mm_xor_si128( casti_m128i( a, 2 ),
-                                        casti_m128i( b, 2 ) );
-   casti_m128i( c, 3 ) = _mm_xor_si128( casti_m128i( a, 3 ),
-                                        casti_m128i( b, 3 ) );
+#elif defined(__SSE2__) || defined(__ARM_NEON)
+   casti_v128( c, 0 ) = v128_xor( casti_v128( a, 0 ),
+                                  casti_v128( b, 0 ) );
+   casti_v128( c, 1 ) = v128_xor( casti_v128( a, 1 ),
+                                  casti_v128( b, 1 ) );
+   casti_v128( c, 2 ) = v128_xor( casti_v128( a, 2 ),
+                                  casti_v128( b, 2 ) );
+   casti_v128( c, 3 ) = v128_xor( casti_v128( a, 3 ),
+                                  casti_v128( b, 3 ) );
 #else
    const unsigned long long *A=a, *B=b;
 	unsigned long long *C=c;

algo/groestl/aes_ni/groestl-intr-aes.h

@@ -60,54 +60,17 @@ static const v128u64_t SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003 };
 
 #if defined(__ARM_NEON)
 
-// No fast shuffle on NEON
-//static const uint32x4_t vmask_d8 = {  3, 1, 2, 0 };  
-static const v128u32_t BLEND_MASK = { 0xffffffff, 0, 0, 0xffffffff };
+static const v128u32_t gr_mask __attribute__ ((aligned (16))) =
+   { 0x03020100, 0x0b0a0908, 0x07060504, 0x0f0e0d0c };
 
-#define gr_shuffle32( v )      v128_blendv( v128_qrev32( v ), v, BLEND_MASK )
-
-/*
-#define TRANSP_MASK \
-     0xd,0x5,0x9,0x1,0xc,0x4,0x8,0x0,0xf,0x7,0xb,0x3,0xe,0x6,0xa,0x2
-#define SUBSH_MASK0 \
-     0xb,0xe,0x1,0x4,0x7,0xa,0xd,0x0,0x3,0x6,0x9,0xc,0xf,0x2,0x5,0x8
-#define SUBSH_MASK1 \
-     0xc,0xf,0x2,0x5,0x8,0xb,0xe,0x1,0x4,0x7,0xa,0xd,0x0,0x3,0x6,0x9
-#define SUBSH_MASK2 \
-     0xd,0x0,0x3,0x6,0x9,0xc,0xf,0x2,0x5,0x8,0xb,0xe,0x1,0x4,0x7,0xa
-#define SUBSH_MASK3 \
-     0xe,0x1,0x4,0x7,0xa,0xd,0x0,0x3,0x6,0x9,0xc,0xf,0x2,0x5,0x8,0xb
-#define SUBSH_MASK4  \
-     0xf,0x2,0x5,0x8,0xb,0xe,0x1,0x4,0x7,0xa,0xd,0x0,0x3,0x6,0x9,0xc
-#define SUBSH_MASK5 \
-     0x0,0x3,0x6,0x9,0xc,0xf,0x2,0x5,0x8,0xb,0xe,0x1,0x4,0x7,0xa,0xd
-#define SUBSH_MASK6 \
-     0x1,0x4,0x7,0xa,0xd,0x0,0x3,0x6,0x9,0xc,0xf,0x2,0x5,0x8,0xb,0xe
-#define SUBSH_MASK7 \
-     0x6,0x9,0xc,0xf,0x2,0x5,0x8,0xb,0xe,0x1,0x4,0x7,0xa,0xd,0x0,0x3
-
-//#define gr_shuffle8( v, c )    v128_shullfev8( v, c )
-
-
-#define gr_shuffle8( v, c15, c14, c13, c12, c11, c10, c09, c08, \
-                        c07, c06, c05, c04, c03, c02, c01, c00 ) \
-  v128_movlane8( v128_movlane8( v128_movlane8( v128_movlane8( \
-  v128_movlane8( v128_movlane8( v128_movlane8( v128_movlane8( \
-  v128_movlane8( v128_movlane8( v128_movlane8( v128_movlane8( \
-  v128_movlane8( v128_movlane8( v128_movlane8( v128_movlane8( \
-    v, 15, v, c15 ), 14, v, c14 ), 13, v, c13 ), 12, v, c12 ), \
-       11, v, c11 ), 10, v, c10 ),  9, v, c09 ),  8, v, c08 ), \
-        7, v, c07 ),  6, v, c06 ),  5, v, c05 ),  4, v, c04 ), \
-        3, v, c03 ),  2, v, c02 ),  1, v, c01 ),  0, v, c00 )
-*/
+#define gr_shuffle32(v)       vqtbl1q_u8( v, gr_mask ) 
 
 #else
 
-#define gr_shuffle32( v )       _mm_shuffle_epi32( v, 0xd8 )
+#define gr_shuffle32(v)       _mm_shuffle_epi32( v, 0xd8 )
 
 #endif
 
-
 #define tos(a)    #a
 #define tostr(a)  tos(a)
 
@@ -140,7 +103,7 @@ static const v128u32_t BLEND_MASK = { 0xffffffff, 0, 0, 0xffffffff };
    This implementation costs 7.7 c/b giving total speed on SNB: 10.7c/b.
    K. Matusiewicz, 2011/05/29 */
 
-#if defined(__AVX512VL__)
+#if defined(VL256)
 
 #define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
   /* t_i = a_i + a_{i+1} */\
@@ -334,17 +297,16 @@ static const v128u32_t BLEND_MASK = { 0xffffffff, 0, 0, 0xffffffff };
  */
 #define SUBMIX(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
   /* SubBytes */\
-  b0 = v128_xor(b0, b0);\
-  a0 = v128_aesenclast(a0, b0);\
-  a1 = v128_aesenclast(a1, b0);\
-  a2 = v128_aesenclast(a2, b0);\
-  a3 = v128_aesenclast(a3, b0);\
-  a4 = v128_aesenclast(a4, b0);\
-  a5 = v128_aesenclast(a5, b0);\
-  a6 = v128_aesenclast(a6, b0);\
-  a7 = v128_aesenclast(a7, b0);\
+  a0 = v128_aesenclast_nokey( a0 ); \
+  a1 = v128_aesenclast_nokey( a1 ); \
+  a2 = v128_aesenclast_nokey( a2 ); \
+  a3 = v128_aesenclast_nokey( a3 ); \
+  a4 = v128_aesenclast_nokey( a4 ); \
+  a5 = v128_aesenclast_nokey( a5 ); \
+  a6 = v128_aesenclast_nokey( a6 ); \
+  a7 = v128_aesenclast_nokey( a7 ); \
   /* MixBytes */\
-  MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7);\
+  MixBytes( a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7 ); \
 }
 
 #define ROUNDS_P(){\
@@ -362,10 +324,9 @@ static const v128u32_t BLEND_MASK = { 0xffffffff, 0, 0, 0xffffffff };
     xmm13 = v128_shuffle8( xmm13, SUBSH_MASK5 ); \
     xmm14 = v128_shuffle8( xmm14, SUBSH_MASK6 ); \
     xmm15 = v128_shuffle8( xmm15, SUBSH_MASK7 ); \
-    /* SubBytes + MixBytes */\
+     /* SubBytes + MixBytes */\
     SUBMIX( xmm8, xmm9, xmm10, xmm11, xmm12, xmm13, xmm14, xmm15, \
             xmm0, xmm1, xmm2,  xmm3,  xmm4,  xmm5,  xmm6,  xmm7 ); \
-    \
     /* AddRoundConstant P1024 */\
     xmm0 = v128_xor( xmm0, \
              casti_v128( round_const_p, round_counter+1 ) ); \
@@ -467,7 +428,6 @@ static const v128u32_t BLEND_MASK = { 0xffffffff, 0, 0, 0xffffffff };
   t1 = v128_unpackhi16(t1, i3);\
   i2 = v128_unpacklo16(i2, i3);\
   i0 = v128_unpacklo16(i0, i1);\
-\
   /* shuffle with immediate */\
   t0 = gr_shuffle32( t0 ); \
   t1 = gr_shuffle32( t1 ); \
@@ -477,7 +437,6 @@ static const v128u32_t BLEND_MASK = { 0xffffffff, 0, 0, 0xffffffff };
   i2 = gr_shuffle32( i2 ); \
   i4 = gr_shuffle32( i4 ); \
   i6 = gr_shuffle32( i6 ); \
-\
   /* continue with unpack */\
   t4 = i0;\
   i0 = v128_unpacklo32(i0, i2);\
@@ -584,7 +543,8 @@ static const v128u32_t BLEND_MASK = { 0xffffffff, 0, 0, 0xffffffff };
   /* transpose done */\
 }/**/
 
-
+#if 0
+// not used
 void INIT( v128_t* chaining )
 {
   static v128_t xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7;
@@ -613,6 +573,7 @@ void INIT( v128_t* chaining )
   chaining[6] = xmm14;
   chaining[7] = xmm15;
 }
+#endif
 
 void TF1024( v128_t* chaining, const v128_t* message )
 {

algo/groestl/aes_ni/groestl256-intr-aes.h

@@ -1,3 +1,6 @@
+#if !defined GROESTL256_INTR_AES_H__
+#define GROESTL256_INTR_AES_H__
+
 /* groestl-intr-aes.h     Aug 2011
  *
  * Groestl implementation with intrinsics using ssse3, sse4.1, and aes
@@ -50,18 +53,17 @@ static const v128u64_t SUBSH_MASK7 = { 0x090c000306080b07, 0x02050f0a0d01040e };
 
 #if defined(__ARM_NEON)
 
-// No fast shuffle on NEON
-static const uint32x4_t vmask_d8 = {  3, 1, 2, 0 };
+static const v128u32_t gr_mask __attribute__ ((aligned (16))) =
+   { 0x03020100, 0x0b0a0908, 0x07060504, 0x0f0e0d0c };
 
-#define gr_shuffle32( v )       v128_shufflev32( v, vmask_d8 )
+#define gr_shuffle32(v)       vqtbl1q_u8( v, gr_mask ) 
 
 #else
 
-#define gr_shuffle32( v )       _mm_shuffle_epi32( v, 0xd8 )
+#define gr_shuffle32(v)       _mm_shuffle_epi32( v, 0xd8 )
 
 #endif
 
-
 #define tos(a)    #a
 #define tostr(a)  tos(a)
 
@@ -93,7 +95,7 @@ static const uint32x4_t vmask_d8 = {  3, 1, 2, 0 };
    This implementation costs 7.7 c/b giving total speed on SNB: 10.7c/b.
    K. Matusiewicz, 2011/05/29 */
 
-#if defined(__AVX512VL__)
+#if defined(VL256)
 
 #define MixBytes(a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
   /* t_i = a_i + a_{i+1} */\
@@ -598,4 +600,4 @@ void OF512( v128_t* chaining )
   chaining[3] = xmm11;
 }
 
-
+#endif

algo/groestl/aes_ni/hash-groestl.c

@@ -146,7 +146,7 @@ int groestl512( hashState_groestl* ctx, void* output, const void* input,
    const int hash_offset = SIZE512 - hashlen_m128i;
    uint64_t blocks = len / SIZE512;
    v128_t* in = (v128_t*)input;
-
+   
    // digest any full blocks, process directly from input 
    for ( i = 0; i < blocks; i++ )
       TF1024( ctx->chaining, &in[ i * SIZE512 ] );
@@ -181,6 +181,7 @@ int groestl512( hashState_groestl* ctx, void* output, const void* input,
 
    // digest final padding block and do output transform
    TF1024( ctx->chaining, ctx->buffer );
+
    OF1024( ctx->chaining );
 
    // store hash result in output 

algo/groestl/aes_ni/hash-groestl.h

@@ -87,6 +87,7 @@ int final_groestl( hashState_groestl*, void* );
 int update_and_final_groestl( hashState_groestl*,  void*, const void*, int );
 int groestl512( hashState_groestl*,  void*, const void*, uint64_t );
 #define groestl512_full   groestl512
+#define groestl512_ctx    groestl512
 
 
 #endif /* __hash_h */

algo/groestl/groestl-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__VAES__) && defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(__VAES__) && defined(SIMD512)
   #define GROESTL_4WAY_VAES 1
 #endif
 

algo/groestl/groestl256-hash-4way.c

@@ -17,7 +17,7 @@
 
 #if defined(__AVX2__) && defined(__VAES__)
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 
 int groestl256_4way_init( groestl256_4way_context* ctx, uint64_t hashlen )

algo/groestl/groestl256-hash-4way.h

@@ -43,7 +43,7 @@
 
 #define SIZE256 (SIZE_512/16)
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct {
   __attribute__ ((aligned (128))) __m512i chaining[SIZE256];

algo/groestl/groestl256-intr-4way.h

@@ -42,7 +42,7 @@ static const __m128i round_const_l7[] __attribute__ ((aligned (64))) =
    { 0x0000000000000000, 0x8696a6b6c6d6e6f6 }
 };
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 static const __m512i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02,
                                      0x1d1519111c141810, 0x1f171b131e161a12,
@@ -626,7 +626,7 @@ static const __m256i SUBSH_MASK7_2WAY =
 
 #define ROUND_2WAY(i, a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3, b4, b5, b6, b7){\
   /* AddRoundConstant */\
-  b1 = mm256_bcast_m128( mm128_mask_32( v128_neg1, 0x3 ) ); \
+  b1 = mm256_bcast_m128( v128_mask32( v128_neg1, 0x3 ) ); \
   a0 = _mm256_xor_si256( a0, mm256_bcast_m128( round_const_l0[i] ) );\
   a1 = _mm256_xor_si256( a1, b1 );\
   a2 = _mm256_xor_si256( a2, b1 );\

algo/groestl/groestl512-hash-4way.c

@@ -17,7 +17,7 @@
 
 #if defined(__AVX2__) && defined(__VAES__)
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 int groestl512_4way_init( groestl512_4way_context* ctx, uint64_t hashlen )
 {

algo/groestl/groestl512-hash-4way.h

@@ -33,7 +33,7 @@
 
 #define SIZE512 (SIZE_1024/16)
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct {
   __attribute__ ((aligned (128))) __m512i chaining[SIZE512];

algo/groestl/groestl512-intr-4way.h

@@ -50,7 +50,7 @@ static const __m128i round_const_q[] __attribute__ ((aligned (64))) =
    { 0x8292a2b2c2d2e2f2, 0x0212223242526272 }
 };
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 static const __m512i TRANSP_MASK = { 0x0d0509010c040800, 0x0f070b030e060a02,
                                      0x1d1519111c141810, 0x1f171b131e161a12,
@@ -239,7 +239,7 @@ static const __m512i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003,
   { \
     /* AddRoundConstant P1024 */\
     xmm8 = _mm512_xor_si512( xmm8, mm512_bcast_m128( \
-             casti_m128i( round_const_p, round_counter ) ) ); \
+             casti_v128u32( round_const_p, round_counter ) ) ); \
     /* ShiftBytes P1024 + pre-AESENCLAST */\
     xmm8  = _mm512_shuffle_epi8( xmm8,  SUBSH_MASK0 ); \
     xmm9  = _mm512_shuffle_epi8( xmm9,  SUBSH_MASK1 );\
@@ -254,7 +254,7 @@ static const __m512i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003,
     \
      /* AddRoundConstant P1024 */\
     xmm0 = _mm512_xor_si512( xmm0, mm512_bcast_m128( \
-             casti_m128i( round_const_p, round_counter+1 ) ) ); \
+             casti_v128u32( round_const_p, round_counter+1 ) ) ); \
     /* ShiftBytes P1024 + pre-AESENCLAST */\
     xmm0 = _mm512_shuffle_epi8( xmm0, SUBSH_MASK0 );\
     xmm1 = _mm512_shuffle_epi8( xmm1, SUBSH_MASK1 );\
@@ -283,7 +283,7 @@ static const __m512i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003,
     xmm13 = _mm512_xor_si512( xmm13, xmm1 );\
     xmm14 = _mm512_xor_si512( xmm14, xmm1 );\
     xmm15 = _mm512_xor_si512( xmm15, mm512_bcast_m128( \
-                 casti_m128i( round_const_q, round_counter ) ) ); \
+                 casti_v128u32( round_const_q, round_counter ) ) ); \
     /* ShiftBytes Q1024 + pre-AESENCLAST */\
     xmm8  = _mm512_shuffle_epi8( xmm8,  SUBSH_MASK1 );\
     xmm9  = _mm512_shuffle_epi8( xmm9,  SUBSH_MASK3 );\
@@ -306,7 +306,7 @@ static const __m512i SUBSH_MASK7 = { 0x06090c0f0205080b, 0x0e0104070a0d0003,
     xmm5 = _mm512_xor_si512( xmm5, xmm9 );\
     xmm6 = _mm512_xor_si512( xmm6, xmm9 );\
     xmm7 = _mm512_xor_si512( xmm7, mm512_bcast_m128( \
-             casti_m128i( round_const_q, round_counter+1 ) ) ); \
+             casti_v128u32( round_const_q, round_counter+1 ) ) ); \
     /* ShiftBytes Q1024 + pre-AESENCLAST */\
     xmm0 = _mm512_shuffle_epi8( xmm0, SUBSH_MASK1 );\
     xmm1 = _mm512_shuffle_epi8( xmm1, SUBSH_MASK3 );\
@@ -812,7 +812,7 @@ static const __m256i SUBSH_MASK7_2WAY =
   { \
     /* AddRoundConstant P1024 */\
     xmm8 = _mm256_xor_si256( xmm8, mm256_bcast_m128( \
-             casti_m128i( round_const_p, round_counter ) ) ); \
+             casti_v128u32( round_const_p, round_counter ) ) ); \
     /* ShiftBytes P1024 + pre-AESENCLAST */\
     xmm8  = _mm256_shuffle_epi8( xmm8,  SUBSH_MASK0_2WAY ); \
     xmm9  = _mm256_shuffle_epi8( xmm9,  SUBSH_MASK1_2WAY );\
@@ -827,7 +827,7 @@ static const __m256i SUBSH_MASK7_2WAY =
     \
      /* AddRoundConstant P1024 */\
     xmm0 = _mm256_xor_si256( xmm0, mm256_bcast_m128( \
-             casti_m128i( round_const_p, round_counter+1 ) ) ); \
+             casti_v128u32( round_const_p, round_counter+1 ) ) ); \
     /* ShiftBytes P1024 + pre-AESENCLAST */\
     xmm0 = _mm256_shuffle_epi8( xmm0, SUBSH_MASK0_2WAY );\
     xmm1 = _mm256_shuffle_epi8( xmm1, SUBSH_MASK1_2WAY );\
@@ -856,7 +856,7 @@ static const __m256i SUBSH_MASK7_2WAY =
     xmm13 = _mm256_xor_si256( xmm13, xmm1 );\
     xmm14 = _mm256_xor_si256( xmm14, xmm1 );\
     xmm15 = _mm256_xor_si256( xmm15, mm256_bcast_m128( \
-                 casti_m128i( round_const_q, round_counter ) ) ); \
+                 casti_v128u32( round_const_q, round_counter ) ) ); \
     /* ShiftBytes Q1024 + pre-AESENCLAST */\
     xmm8  = _mm256_shuffle_epi8( xmm8,  SUBSH_MASK1_2WAY );\
     xmm9  = _mm256_shuffle_epi8( xmm9,  SUBSH_MASK3_2WAY );\
@@ -879,7 +879,7 @@ static const __m256i SUBSH_MASK7_2WAY =
     xmm5 = _mm256_xor_si256( xmm5, xmm9 );\
     xmm6 = _mm256_xor_si256( xmm6, xmm9 );\
     xmm7 = _mm256_xor_si256( xmm7, mm256_bcast_m128( \
-             casti_m128i( round_const_q, round_counter+1 ) ) ); \
+             casti_v128u32( round_const_q, round_counter+1 ) ) ); \
     /* ShiftBytes Q1024 + pre-AESENCLAST */\
     xmm0 = _mm256_shuffle_epi8( xmm0, SUBSH_MASK1_2WAY );\
     xmm1 = _mm256_shuffle_epi8( xmm1, SUBSH_MASK3_2WAY );\

algo/groestl/myrgr-4way.c

@@ -213,7 +213,7 @@ int scanhash_myriad_4way( struct work *work, uint32_t max_nonce,
 
    v128_bswap32_intrlv80_4x32( vdata, pdata );
    do {
-      *noncev = mm128_bswap_32( _mm_set_epi32( n+3,n+2,n+1,n ) );
+      *noncev = v128_bswap32( _mm_set_epi32( n+3,n+2,n+1,n ) );
 
       myriad_4way_hash( hash, vdata );
       pdata[19] = n;

algo/groestl/myrgr-gate.c

@@ -16,7 +16,7 @@ bool register_myriad_algo( algo_gate_t* gate )
   init_myrgr_ctx();
   gate->scanhash  = (void*)&scanhash_myriad;
   gate->hash      = (void*)&myriad_hash;
-  gate->optimizations = AES_OPT | SSE2_OPT | AVX2_OPT | SHA_OPT | VAES_OPT;
+  gate->optimizations = AES_OPT | SSE2_OPT | AVX2_OPT | SHA256_OPT | VAES_OPT;
 #endif
   return true;
 };

algo/groestl/myrgr-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__VAES__) && defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(__VAES__) && defined(SIMD512)
   #define MYRGR_8WAY 1
 #elif defined(__AVX2__) && defined(__AES__) && !defined(__SHA__)
   #define MYRGR_4WAY 1

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

@@ -382,12 +382,12 @@ static const uint32_t T512[64][16] __attribute__ ((aligned (32))) =
 #define S1F   MF
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // Hamsi 8 way AVX512 
 
 // Intel docs say _mm512_movepi64_mask & _mm512_cmplt_epi64_mask have same
-// timig. However, when tested hashing X13 on i9-9940x using cmplt with zero
+// timing. However, testing hashing X13 on i9-9940x using cmplt with zero
 // had a 3% faster overall hashrate than than using movepi. 
 
 #define INPUT_BIG8 \
@@ -418,13 +418,11 @@ static const uint32_t T512[64][16] __attribute__ ((aligned (32))) =
   tb = mm512_xoror( b, d, a ); \
   a = _mm512_xor_si512( a, c ); \
   b = mm512_xoror( td, tb, a ); \
-  td = mm512_xorand( a, td, tb ); \
+  d = _mm512_ternarylogic_epi64( a, td, tb, 0x87 );/* not( xorand( a, td, tb ) ); */ \
   a = c; \
-  c = mm512_xor3( tb, b, td ); \
-  d = mm512_not( td ); \
+  c = _mm512_ternarylogic_epi64( tb, b, d, 0x69 ); /* not( xor3( tb, b, d ) ); */ \
 }
 
-
 /*
 #define SBOX8( a, b, c, d ) \
 do { \
@@ -1122,7 +1120,7 @@ void hamsi512_8way_close( hamsi_8way_big_context *sc, void *dst )
 
 // Hamsi 4 way AVX2
 
-#if defined(__AVX512VL__)
+#if defined(VL256)
 
 #define INPUT_BIG \
 do { \
@@ -1155,11 +1153,99 @@ do { \
   b = mm256_xoror( td, tb, a ); \
   d = _mm256_ternarylogic_epi64( a, td, tb, 0x87 );/* mm256_not( mm256_xorand( a, td, tb ) ); */ \
   a = c; \
-  c = _mm256_ternarylogic_epi64( tb, b, d, 0x69 ); /*mm256_not( mm256_xor3( tb, b, d ) );*/ \
+  c = _mm256_ternarylogic_epi64( tb, b, d, 0x69 ); /* mm256_not( mm256_xor3( tb, b, d ) ); */ \
 }
 
 #else
 
+#define INPUT_BIG_sub( db_i ) \
+{ \
+     const __m256i dm = _mm256_cmpgt_epi64( zero, db_i ); \
+     m0 = _mm256_xor_si256( m0, _mm256_and_si256( dm, v256_64( tp[0] ) ) ); \
+     m1 = _mm256_xor_si256( m1, _mm256_and_si256( dm, v256_64( tp[1] ) ) ); \
+     m2 = _mm256_xor_si256( m2, _mm256_and_si256( dm, v256_64( tp[2] ) ) ); \
+     m3 = _mm256_xor_si256( m3, _mm256_and_si256( dm, v256_64( tp[3] ) ) ); \
+     m4 = _mm256_xor_si256( m4, _mm256_and_si256( dm, v256_64( tp[4] ) ) ); \
+     m5 = _mm256_xor_si256( m5, _mm256_and_si256( dm, v256_64( tp[5] ) ) ); \
+     m6 = _mm256_xor_si256( m6, _mm256_and_si256( dm, v256_64( tp[6] ) ) ); \
+     m7 = _mm256_xor_si256( m7, _mm256_and_si256( dm, v256_64( tp[7] ) ) ); \
+     tp += 8; \
+}
+
+#define INPUT_BIG \
+{ \
+  const __m256i db = *buf; \
+  const __m256i zero = m256_zero; \
+  const uint64_t *tp = (const uint64_t*)T512;  \
+  m0 = m1 = m2 = m3 = m4 = m5 = m6 = m7 = zero; \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,63 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,62 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,61 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,60 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,59 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,58 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,57 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,56 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,55 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,54 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,53 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,52 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,51 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,50 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,49 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,48 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,47 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,46 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,45 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,44 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,43 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,42 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,41 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,40 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,39 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,38 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,37 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,36 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,35 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,34 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,33 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,32 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,31 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,30 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,29 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,28 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,27 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,26 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,25 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,24 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,23 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,22 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,21 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,20 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,19 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,18 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,17 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,16 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,15 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,14 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,13 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,12 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,11 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db,10 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 9 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 8 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 7 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 6 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 5 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 4 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 3 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 2 ) ); \
+  INPUT_BIG_sub( _mm256_slli_epi64( db, 1 ) ); \
+  INPUT_BIG_sub( db ); \
+}
+
+#if 0
+// dependent on the compiler unrolling the loop
 #define INPUT_BIG \
 do { \
   __m256i db = *buf; \
@@ -1180,6 +1266,7 @@ do { \
      tp += 8; \
   } \
 } while (0)
+#endif
 
 // v3 no ternary logic, 15 instructions, 9 TL equivalent instructions
 #define SBOX( a, b, c, d ) \
@@ -1219,7 +1306,7 @@ do { \
 do { \
    a = mm256_rol_32( a, 13 ); \
    c = mm256_rol_32( c,  3 ); \
-   b = mm256_xor3( a, b, c ); \
+   b = mm256_xor3( b, a, c ); \
    d = mm256_xor3( d, c, _mm256_slli_epi32( a, 3 ) ); \
    b = mm256_rol_32( b, 1 ); \
    d = mm256_rol_32( d, 7 ); \
@@ -1501,7 +1588,7 @@ do { /* order is important */ \
    sc->h[14] = CE; \
    sc->h[15] = CF;
 
-#if defined(__AVX512VL__)
+#if defined(VL256)
 
 #define INPUT_8X32 \
 { \
@@ -1961,6 +2048,94 @@ void hamsi512_4way_close( hamsi_4way_big_context *sc, void *dst )
    sc->h[6] = c6; \
    sc->h[7] = c7;
 
+#define INPUT_2x64_sub( db_i ) \
+{ \
+     const v128u64_t dm = v128_cmpgt64( zero, db_i ); \
+     m0 = v128_xor( m0, v128_and( dm, v128_64( tp[0] ) ) ); \
+     m1 = v128_xor( m1, v128_and( dm, v128_64( tp[1] ) ) ); \
+     m2 = v128_xor( m2, v128_and( dm, v128_64( tp[2] ) ) ); \
+     m3 = v128_xor( m3, v128_and( dm, v128_64( tp[3] ) ) ); \
+     m4 = v128_xor( m4, v128_and( dm, v128_64( tp[4] ) ) ); \
+     m5 = v128_xor( m5, v128_and( dm, v128_64( tp[5] ) ) ); \
+     m6 = v128_xor( m6, v128_and( dm, v128_64( tp[6] ) ) ); \
+     m7 = v128_xor( m7, v128_and( dm, v128_64( tp[7] ) ) ); \
+     tp += 8; \
+}
+
+#define INPUT_2x64 \
+{ \
+  const v128u64_t db = *buf; \
+  const v128u64_t zero = v128_zero; \
+  const uint64_t *tp = (const uint64_t*)T512;  \
+  m0 = m1 = m2 = m3 = m4 = m5 = m6 = m7 = zero; \
+  INPUT_2x64_sub( v128_sl64( db,63 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,62 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,61 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,60 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,59 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,58 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,57 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,56 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,55 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,54 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,53 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,52 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,51 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,50 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,49 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,48 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,47 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,46 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,45 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,44 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,43 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,42 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,41 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,40 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,39 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,38 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,37 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,36 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,35 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,34 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,33 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,32 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,31 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,30 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,29 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,28 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,27 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,26 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,25 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,24 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,23 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,22 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,21 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,20 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,19 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,18 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,17 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,16 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,15 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,14 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,13 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,12 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,11 ) ); \
+  INPUT_2x64_sub( v128_sl64( db,10 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 9 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 8 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 7 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 6 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 5 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 4 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 3 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 2 ) ); \
+  INPUT_2x64_sub( v128_sl64( db, 1 ) ); \
+  INPUT_2x64_sub( db ); \
+}
+
+#if 0
+// Dependent on the compiler unrolling the loop.
 #define INPUT_2x64 \
 { \
   v128u64_t db = *buf; \
@@ -1981,6 +2156,7 @@ void hamsi512_4way_close( hamsi_4way_big_context *sc, void *dst )
      tp += 8; \
   } \
 }
+#endif
 
 // v3 no ternary logic, 15 instructions, 9 TL equivalent instructions
 #define SBOX_2x64( a, b, c, d ) \
@@ -2001,7 +2177,7 @@ void hamsi512_4way_close( hamsi_4way_big_context *sc, void *dst )
 { \
    a = v128_rol32( a, 13 ); \
    c = v128_rol32( c,  3 ); \
-   b = v128_xor3( a, b, c ); \
+   b = v128_xor3( c, a, b ); \
    d = v128_xor3( d, c, v128_sl32( a, 3 ) ); \
    b = v128_rol32( b, 1 ); \
    d = v128_rol32( d, 7 ); \

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

@@ -104,7 +104,7 @@ void hamsi512_8x32_full( hamsi512_8x32_context *sc, void *dst, const void *data,
 
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // Hamsi-512 8x64
 

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

@@ -53,7 +53,7 @@ extern "C"{
 #define SPH_SMALL_FOOTPRINT_HAVAL   1
 //#endif
 
-#if defined(__AVX512VL__)
+#if defined(VL256)
 
 // ( ~( a ^ b ) ) & c
 #define v128_andnotxor( a, b, c ) \
@@ -583,7 +583,7 @@ do { \
 
 // Haval-256 8 way 32 bit avx2
 
-#if defined (__AVX512VL__)
+#if defined (VL256)
 
 // ( ~( a ^ b ) ) & c
 #define mm256_andnotxor( a, b, c ) \
@@ -882,7 +882,7 @@ do { \
 
 #endif // AVX2
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512) 
 
 // ( ~( a ^ b ) ) & c
 #define mm512_andnotxor( a, b, c ) \

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

@@ -107,7 +107,7 @@ void haval256_5_8way_close( void *cc, void *dst );
 
 #endif // AVX2
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct {
    __m512i buf[32];

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

@@ -204,7 +204,7 @@ static const uint64_t IV512[] =
       (state)->H[15] = h7l; \
    } while (0)
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 #define Sb_8W(x0, x1, x2, x3, c) \
 { \
@@ -364,8 +364,7 @@ static const uint64_t IV512[] =
 
 #if defined(__AVX2__)
 
-#if defined(__AVX512VL__)
-//TODO enable for AVX10_256, not used with AVX512VL
+#if defined(VL256)
 
 #define notxorandnot( a, b, c ) \
    _mm256_ternarylogic_epi64( a, b, c, 0x2d )
@@ -522,7 +521,7 @@ static const uint64_t IV512[] =
 
 #endif   // AVX2
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 void jh256_8x64_init( jh_8x64_context *sc )
 {

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

@@ -55,7 +55,7 @@
  * <code>memcpy()</code>).
  */
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct
 {

algo/keccak/keccak-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define KECCAK_8WAY 1
 #elif defined(__AVX2__)
   #define KECCAK_4WAY 1
@@ -12,7 +12,7 @@
   #define KECCAK_2WAY 1
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define SHA3D_8WAY 1
 #elif defined(__AVX2__)
   #define SHA3D_4WAY 1

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

@@ -57,7 +57,7 @@ static const uint64_t RC[] = {
 
 #define DO(x)   x
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 #define INPUT_BUF(size)   do { \
     size_t j; \

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

@@ -4,7 +4,7 @@
 #include <stddef.h>
 #include "simd-utils.h"
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct
 {

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

@@ -59,7 +59,7 @@ static const uint32_t CNS_INIT[128] __attribute((aligned(64))) = {
 };
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 #define cns4w(i)  mm512_bcast_m128( ( (__m128i*)CNS_INIT)[i] )
 
@@ -524,8 +524,7 @@ int luffa_4way_update_close( luffa_4way_context *state,
     a = _mm256_xor_si256( a, c0 ); \
     b = _mm256_xor_si256( b, c1 );
 
-//TODO Enable for AVX10_256, not used with AVX512 or AVX10_512
-#if defined(__AVX512VL__) 
+#if defined(VL256) 
 
 #define MULT2( a0, a1 ) \
 { \

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

@@ -51,7 +51,7 @@
 #define LIMIT_512 128
 /*********************************/
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct {
     uint32_t buffer[8*4];

algo/luffa/luffa_for_sse2.c

@@ -28,8 +28,7 @@
     a = v128_xor( a, c0 ); \
     b = v128_xor( b, c1 ); \
 
-#if defined(__AVX512VL__)
-//TODO enable for AVX10_512 AVX10_256
+#if defined(VL256)
 
 #define MULT2( a0, a1 ) \
 { \
@@ -48,29 +47,22 @@
   a1 = _mm_alignr_epi8( b, a1, 4 ); \
 }
 
-#elif defined(__ARM_NEON)
+
+#elif defined(__ARM_NEON) || defined(__SSE2__)
 
 // { a1_0, 0, a1_0, a1_0 }
 #define MULT2( a0, a1 ) \
 { \
-  v128_t b = v128_xor( a0, v128_and( vdupq_laneq_u32( a1, 0 ), MASK ) ); \
+  v128_t b = v128_xor( a0, v128_and( v128_bcast32( a1 ), MASK ) ); \
   a0 = v128_alignr32( a1, b, 1 ); \
   a1 = v128_alignr32( b, a1, 1 ); \
 }
 
-#else   // assume SSE2
-
-#define MULT2( a0, a1 ) \
-{ \
-  v128_t b = v128_xor( a0, v128_and( _mm_shuffle_epi32( a1, 0 ), MASK ) ); \
-  a0 = v128_or( _mm_srli_si128(  b, 4 ), _mm_slli_si128( a1, 12 ) ); \
-  a1 = v128_or( _mm_srli_si128( a1, 4 ), _mm_slli_si128(  b, 12 ) ); \
-} 
-
+#else
+  #warning __FILE__ ":" __LINE__ " Unknown or unsupported CPU architecture."
 #endif
 
-#if defined(__AVX512VL__)
-//TODO enable for AVX10_512 AVX10_256
+#if defined(VL256)
 
 #define SUBCRUMB( a0, a1, a2, a3 ) \
 { \

algo/luffa/luffa_for_sse2.h

@@ -68,4 +68,4 @@ int update_and_final_luffa( hashState_luffa *state, void* output,
 
 int luffa_full( hashState_luffa *state, void* output, int hashbitlen,
                                    const void* data, size_t inlen );
-#endif   // LUFFA_FOR_SSE2_H___
+#endif   // LUFFA_FOR_SSE2_H__

algo/lyra2/allium-4way.c

@@ -15,7 +15,7 @@
  #include "algo/groestl/sph_groestl.h"
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define ALLIUM_16WAY 1
 #elif defined(__AVX2__)
   #define ALLIUM_8WAY 1
@@ -465,12 +465,8 @@ typedef union
 {
    keccak256_2x64_context    keccak;
    cubehashParam             cube;
-//#if defined(__x86_64__)
    skein256_2x64_context     skein;
-//#else
-//   sph_skein512_context      skein;
-//#endif
-#if defined(__AES__) // || defined(__ARM_FEATURE_AES)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
    hashState_groestl256      groestl;
 #else
    sph_groestl256_context     groestl;
@@ -516,7 +512,6 @@ static void allium_4way_hash( void *hash, const void *midstate_vars,
    LYRA2RE( hash2, 32, hash2, 32, hash2, 32, 1, 8, 8 );
    LYRA2RE( hash3, 32, hash3, 32, hash3, 32, 1, 8, 8 );
 
-//#if defined(__x86_64__)
    intrlv_2x64( vhashA, hash0, hash1, 256 );
    skein256_2x64_init( &ctx.skein );
    skein256_2x64_update( &ctx.skein, vhashA, 32 );
@@ -527,23 +522,8 @@ static void allium_4way_hash( void *hash, const void *midstate_vars,
    skein256_2x64_update( &ctx.skein, vhashA, 32 );
    skein256_2x64_close( &ctx.skein, vhashA );
    dintrlv_2x64( hash2, hash3, vhashA, 256 );
-/*
-#else
-    sph_skein256_init( &ctx.skein );
-    sph_skein256( &ctx.skein, hash0, 32 );
-    sph_skein256_close( &ctx.skein, hash0 );
-    sph_skein256_init( &ctx.skein );
-    sph_skein256( &ctx.skein, hash1, 32 );
-    sph_skein256_close( &ctx.skein, hash1 );
-    sph_skein256_init( &ctx.skein );
-    sph_skein256( &ctx.skein, hash2, 32 );
-    sph_skein256_close( &ctx.skein, hash2 );
-    sph_skein256_init( &ctx.skein );
-    sph_skein256( &ctx.skein, hash3, 32 );
-    sph_skein256_close( &ctx.skein, hash3 );
-#endif
-*/
-#if defined(__AES__) // || defined(__ARM_FEATURE_AES)
+
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
    groestl256_full( &ctx.groestl, hash0, hash0, 256 );
    groestl256_full( &ctx.groestl, hash1, hash1, 256 );
    groestl256_full( &ctx.groestl, hash2, hash2, 256 );

algo/lyra2/lyra2-gate.h

@@ -5,7 +5,7 @@
 #include <stdint.h>
 #include "lyra2.h"
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define LYRA2REV3_16WAY 1
 #elif defined(__AVX2__)
   #define LYRA2REV3_8WAY 1
@@ -49,7 +49,7 @@ bool init_lyra2rev3_ctx();
 
 //////////////////////////////////
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define LYRA2REV2_16WAY 1
 #elif defined(__AVX2__)
   #define LYRA2REV2_8WAY 1
@@ -108,7 +108,7 @@ bool lyra2h_thread_init();
 
 /////////////////////////////////////////
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define PHI2_8WAY 1
 #elif defined(__AVX2__) && defined(__AES__)
   #define PHI2_4WAY 1

algo/lyra2/lyra2-hash-2way.c

@@ -41,7 +41,7 @@
 //  lyra2z330, lyra2h, 
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 /**
  * Executes Lyra2 based on the G function from Blake2b. This version supports salts and passwords

algo/lyra2/lyra2.h

@@ -59,7 +59,7 @@ int LYRA2Z( uint64_t*, void *K, uint64_t kLen, const void *pwd,
 
 int LYRA2(void *K, int64_t kLen, const void *pwd, int32_t pwdlen, const void *salt, int32_t saltlen, int64_t timeCost, const int16_t nRows, const int16_t nCols);
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 int LYRA2RE_2WAY( void *K, uint64_t kLen, const void *pwd, uint64_t pwdlen,
                   uint64_t timeCost, uint64_t nRows, uint64_t nCols );

algo/lyra2/lyra2h-4way.c

@@ -67,7 +67,7 @@ int scanhash_lyra2h_4way( struct work *work, uint32_t max_nonce,
    lyra2h_4way_midstate( vdata );
 
    do {
-     *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
+     *noncev = v128_bswap32( _mm_set_epi32( n+3, n+2, n+1, n ) );
       lyra2h_4way_hash( hash, vdata );
 
       for ( int i = 0; i < 4; i++ )

algo/lyra2/lyra2rev2-4way.c

@@ -456,7 +456,7 @@ int scanhash_lyra2rev2_4way( struct work *work, uint32_t max_nonce,
 
    do
    {
-      *noncev = mm128_bswap_32( _mm_set_epi32( n+3, n+2, n+1, n ) );
+      *noncev = v128_bswap32( _mm_set_epi32( n+3, n+2, n+1, n ) );
 
       lyra2rev2_4way_hash( hash, vdata );
 

algo/lyra2/lyra2z-4way.c

@@ -3,7 +3,7 @@
 #include "lyra2.h"
 #include "algo/blake/blake256-hash.h"
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define LYRA2Z_16WAY 1
 #elif defined(__AVX2__)
   #define LYRA2Z_8WAY 1

algo/lyra2/phi2-4way.c

@@ -4,7 +4,7 @@
 #include "algo/gost/sph_gost.h"
 #include "algo/cubehash/cubehash_sse2.h"
 #include "lyra2.h"
-#if defined(__VAES__) && defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(__VAES__) && defined(SIMD512)
   #include "algo/echo/echo-hash-4way.h"
 #elif defined(__AES__)
   #include "algo/echo/aes_ni/hash_api.h"

algo/lyra2/sponge-2way.c

@@ -27,7 +27,7 @@
 #include "lyra2.h"
 #include "simd-utils.h"
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 inline void squeeze_2way( uint64_t *State, byte *Out, unsigned int len )
 {

algo/lyra2/sponge.h

@@ -43,7 +43,7 @@ static const uint64_t blake2b_IV[8] =
   0x1f83d9abfb41bd6bULL, 0x5be0cd19137e2179ULL
 };
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 #define G2W_4X64(a,b,c,d) \
    a = _mm512_add_epi64( a, b ); \
@@ -150,13 +150,13 @@ static const uint64_t blake2b_IV[8] =
 // returns void, all args updated
 #define G_2X64(a,b,c,d) \
    a = v128_add64( a, b ); \
-   d = v128_ror64( v128_xor( d, a), 32 ); \
+   d = v128_ror64xor( d, a, 32 ); \
    c = v128_add64( c, d ); \
-   b = v128_ror64( v128_xor( b, c ), 24 ); \
+   b = v128_ror64xor( b, c, 24 ); \
    a = v128_add64( a, b ); \
-   d = v128_ror64( v128_xor( d, a ), 16 ); \
+   d = v128_ror64xor( d, a, 16 ); \
    c = v128_add64( c, d ); \
-   b = v128_ror64( v128_xor( b, c ), 63 );
+   b = v128_ror64xor( b, c, 63 );
 
 #define LYRA_ROUND_AVX(s0,s1,s2,s3,s4,s5,s6,s7) \
 { \
@@ -195,10 +195,6 @@ static const uint64_t blake2b_IV[8] =
 
 #endif // AVX2 else SSE2
 
-static inline uint64_t rotr64( const uint64_t w, const unsigned c ){
-    return ( w >> c ) | ( w << ( 64 - c ) );
-}
-
 #define G( r, i, a, b, c, d ) \
 { \
     a = a + b; \
@@ -222,7 +218,7 @@ static inline uint64_t rotr64( const uint64_t w, const unsigned c ){
     G( r, 7, v[ 3], v[ 4], v[ 9], v[14] );
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 union _ovly_512
 {

algo/m7m/m7m.c

@@ -21,7 +21,7 @@
 #define EPS1 DBL_EPSILON
 #define EPS2 3.0e-11
 
-inline double exp_n( double xt )
+static inline double exp_n( double xt )
 {
     if ( xt < -700.0 )
         return 0;
@@ -33,7 +33,7 @@ inline double exp_n( double xt )
         return exp( xt );
 }
 
-inline double exp_n2( double x1, double x2 )
+static inline double exp_n2( double x1, double x2 )
 {
     double p1 = -700., p2 = -37., p3 = -0.8e-8, p4 = 0.8e-8,
            p5 = 37., p6 = 700.;
@@ -306,7 +306,7 @@ bool register_m7m_algo( algo_gate_t *gate )
   applog( LOG_ERR, "M7M algo is not supported on MacOS");
   return false;
 #else  
-  gate->optimizations = SHA_OPT;
+  gate->optimizations = SHA256_OPT;
   init_m7m_ctx();
   gate->scanhash              = (void*)&scanhash_m7m_hash;
   gate->build_stratum_request = (void*)&std_be_build_stratum_request;

algo/m7m/magimath.cpp

@@ -1,75 +0,0 @@
-// Copyright (c) 2014 The Magi developers
-// Distributed under the MIT/X11 software license, see the accompanying
-// file COPYING or http://www.opensource.org/licenses/mit-license.php.
-
-#include <iostream>
-#include <cfloat>
-#include <limits>
-#include <math.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdint.h>
-
-#include "magimath.h"
-
-#define EPS1 (std::numeric_limits<double>::epsilon())
-#define EPS2 3.0e-11
-
-static void gauleg(double x1, double x2, double x[], double w[], const int n)
-{
-	int m,j,i;
-	double z1, z, xm, xl, pp, p3, p2, p1;
-	m=(n+1)/2;
-	xm=0.5*(x2+x1);
-	xl=0.5*(x2-x1);
-	for (i=1;i<=m;i++) {
-		z=cos(3.141592654*(i-0.25)/(n+0.5));
-		do {
-			p1=1.0;
-			p2=0.0;
-			for (j=1;j<=n;j++) {
-				p3=p2;
-				p2=p1;
-				p1=((2.0*j-1.0)*z*p2-(j-1.0)*p3)/j;
-			}
-			pp=n*(z*p1-p2)/(z*z-1.0);
-			z1=z;
-			z=z1-p1/pp;
-		} while (fabs(z-z1) > EPS2);
-		x[i]=xm-xl*z;
-		x[n+1-i]=xm+xl*z;
-		w[i]=2.0*xl/((1.0-z*z)*pp*pp);
-		w[n+1-i]=w[i];
-	}
-}
-
-static double GaussianQuad_N(double func(const double), const double a2, const double b2, const int NptGQ)
-{
-	double s=0.0;
-#ifdef _MSC_VER
-#define SW_DIVS 23
-	double x[SW_DIVS+1], w[SW_DIVS+1];
-#else
-	double x[NptGQ+1], w[NptGQ+1];
-#endif
-
-	gauleg(a2, b2, x, w, NptGQ);
-
-	for (int j=1; j<=NptGQ; j++) {
-		s += w[j]*func(x[j]);
-	}
-
-	return s;
-}
-
-static double swit_(double wvnmb)
-{
-	return pow( (5.55243*(exp_n(-0.3*wvnmb/15.762) - exp_n(-0.6*wvnmb/15.762)))*wvnmb, 0.5)
-		/ 1034.66 * pow(sin(wvnmb/65.), 2.);
-}
-
-uint32_t sw_(int nnounce, int divs)
-{
-	double wmax = ((sqrt((double)(nnounce))*(1.+EPS1))/450+100);
-	return ((uint32_t)(GaussianQuad_N(swit_, 0., wmax, divs)*(1.+EPS1)*1.e6));
-}

algo/m7m/magimath.h

@@ -1,54 +0,0 @@
-// Copyright (c) 2014 The Magi developers
-// Distributed under the MIT/X11 software license, see the accompanying
-// file COPYING or http://www.opensource.org/licenses/mit-license.php.
-#ifndef MAGI_MATH_H
-#define MAGI_MATH_H
-
-#include <math.h>
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-uint32_t sw_(int nnounce, int divs);
-
-#ifdef __cplusplus
-}
-#endif
-
-
-inline double exp_n(double xt)
-{
-	double p1 = -700.0, p3 = -0.8e-8, p4 = 0.8e-8, p6 = 700.0;
-	if(xt < p1)
-		return 0;
-	else if(xt > p6)
-		return 1e200;
-	else if(xt > p3 && xt < p4)
-		return (1.0 + xt);
-	else
-		return exp(xt);
-}
-
-// 1 / (1 + exp(x1-x2))
-inline double exp_n2(double x1, double x2)
-{
-	double p1 = -700., p2 = -37., p3 = -0.8e-8, p4 = 0.8e-8, p5 = 37., p6 = 700.;
-	double xt = x1 - x2;
-	if (xt < p1+1.e-200)
-		return 1.;
-	else if (xt > p1 && xt < p2 + 1.e-200)
-		return ( 1. - exp(xt) );
-	else if (xt > p2 && xt < p3 + 1.e-200)
-		return ( 1. / (1. + exp(xt)) );
-	else if (xt > p3 && xt < p4)
-		return ( 1. / (2. + xt) );
-	else if (xt > p4 - 1.e-200 && xt < p5)
-		return ( exp(-xt) / (1. + exp(-xt)) );
-	else if (xt > p5 - 1.e-200 && xt < p6)
-		return ( exp(-xt) );
-	else //if (xt > p6 - 1.e-200)
-		return 0.;
-}
-
-#endif

algo/nist5/nist5-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define NIST5_8WAY 1
 #elif defined(__AVX2__) && defined(__AES__)
   #define NIST5_4WAY 1

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

@@ -71,8 +71,7 @@ do { \
 } while (0)
 
 #define GAMMA_4W(n0, n1, n2, n4)   \
-   (g ## n0 = v128_xor( a ## n0, \
-                             v128_or( a ## n1, v128_not( a ## n2 ) ) ) )
+   (g ## n0 = v128_xor( a ## n0, v128_ornot( a ## n2, a ## n1 ) ) )
 
 #define PI_ALL_4W   do { \
       a0  = g0; \
@@ -312,7 +311,7 @@ do { \
       BUPDATE1_8W( 7, 1 ); \
 } while (0)
 
-#if defined(__AVX512VL__)
+#if defined(VL256)
 
 #define GAMMA_8W(n0, n1, n2, n4)   \
    ( g ## n0 = _mm256_ternarylogic_epi32( a ## n0, a ## n2, a ## n1, 0x4b ) )  

algo/quark/anime-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define ANIME_8WAY 1
 #elif defined(__AVX2__) && defined(__AES__)
   #define ANIME_4WAY 1

algo/quark/hmq1725-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define HMQ1725_8WAY 1
 #elif defined(__AVX2__) && defined(__AES__)
   #define HMQ1725_4WAY 1

algo/quark/hmq1725.c

@@ -6,16 +6,16 @@
 #include <stdint.h>
 #include "algo/blake/blake512-hash.h"
 #include "algo/bmw/sph_bmw.h"
-#if defined(__AES__)
-  #include "algo/groestl/aes_ni/hash-groestl.h"
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
   #include "algo/fugue/fugue-aesni.h"
 #else
-  #include "algo/groestl/sph_groestl.h"
   #include "algo/fugue/sph_fugue.h"
 #endif
 #if defined(__AES__) || defined(__ARM_FEATURE_AES)
+  #include "algo/groestl/aes_ni/hash-groestl.h"
   #include "algo/echo/aes_ni/hash_api.h"
 #else
+  #include "algo/groestl/sph_groestl.h"
   #include "algo/echo/sph_echo.h"
 #endif
 #include "algo/jh/sph_jh.h"
@@ -33,18 +33,18 @@
 
 union _hmq1725_ctx_holder
 {
-   blake512_context    blake;
+   blake512_context        blake;
    sph_bmw512_context      bmw;
-#if defined(__AES__)
-   hashState_groestl       groestl;
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
    hashState_fugue         fugue;
 #else
-   sph_groestl512_context  groestl;
    sph_fugue512_context    fugue;
 #endif
 #if defined(__AES__) || defined(__ARM_FEATURE_AES)
+   hashState_groestl       groestl;
    hashState_echo          echo;
 #else
+   sph_groestl512_context  groestl;
    sph_echo512_context     echo;
 #endif
    sph_skein512_context    skein;
@@ -62,9 +62,6 @@ union _hmq1725_ctx_holder
 };
 typedef union _hmq1725_ctx_holder hmq1725_ctx_holder;
 
-//static hmq1725_ctx_holder hmq1725_ctx __attribute__ ((aligned (64)));
-//static __thread sph_bmw512_context hmq_bmw_mid __attribute__ ((aligned (64)));
-
 extern void hmq1725hash(void *state, const void *input)
 {
     const uint32_t mask = 24;
@@ -82,7 +79,7 @@ extern void hmq1725hash(void *state, const void *input)
 
     if ( hashB[0] & mask )   //1
     {
-#if defined(__AES__)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
        groestl512_full( &ctx.groestl, hashA, hashB, 512 );
 #else
        sph_groestl512_init( &ctx.groestl );
@@ -180,7 +177,7 @@ extern void hmq1725hash(void *state, const void *input)
     sph_hamsi512( &ctx.hamsi, hashA, 64 ); //3
     sph_hamsi512_close( &ctx.hamsi, hashB ); //4
 
-#if defined(__AES__)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
     fugue512_full( &ctx.fugue, hashA, hashB, 64 );
 #else
     sph_fugue512_init( &ctx.fugue );
@@ -211,7 +208,7 @@ extern void hmq1725hash(void *state, const void *input)
 
     if ( hashB[0] & mask ) //7
     {
-#if defined(__AES__)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
        fugue512_full( &ctx.fugue, hashA, hashB, 64 );
 #else
        sph_fugue512_init( &ctx.fugue );
@@ -226,7 +223,7 @@ extern void hmq1725hash(void *state, const void *input)
        sph_sha512_close( &ctx.sha, hashA );
     }
 
-#if defined(__AES__)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
     groestl512_full( &ctx.groestl, hashB, hashA, 512 );
 #else
     sph_groestl512_init( &ctx.groestl );
@@ -262,30 +259,18 @@ extern void hmq1725hash(void *state, const void *input)
 int scanhash_hmq1725( struct work *work, uint32_t max_nonce,
                       uint64_t *hashes_done, struct thr_info *mythr )
 {
-//        uint32_t endiandata[32] __attribute__((aligned(64)));
-        uint32_t endiandata[20] __attribute__((aligned(32)));
-        uint32_t hash64[8] __attribute__((aligned(32)));
-        uint32_t *pdata = work->data;
-        uint32_t *ptarget = work->target;
+   uint32_t endiandata[20] __attribute__((aligned(32)));
+   uint32_t hash64[8] __attribute__((aligned(32)));
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
 	uint32_t n = pdata[19] - 1;
 	const uint32_t first_nonce = pdata[19];
    int thr_id = mythr->id;  // thr_id arg is deprecated
-	//const uint32_t Htarg = ptarget[7];
 
 	//we need bigendian data...
-//        for (int k = 0; k < 32; k++)
-        for (int k = 0; k < 20; k++)
-                be32enc(&endiandata[k], pdata[k]);
+   for (int k = 0; k < 20; k++)
+         be32enc(&endiandata[k], pdata[k]);
 
-//        hmq_bmw512_midstate( endiandata );
-
-//	if (opt_debug) 
-//	{
-//		applog(LOG_DEBUG, "Thr: %02d, firstN: %08x, maxN: %08x, ToDo: %d", thr_id, first_nonce, max_nonce, max_nonce-first_nonce);
-//	}
-	
-	/* I'm to lazy to put the loop in an inline function... so dirty copy'n'paste.... */
-	/* i know that i could set a variable, but i don't know how the compiler will optimize it, not that then the cpu needs to load the value *everytime* in a register */
 	if (ptarget[7]==0) {
 		do {
 			pdata[19] = ++n;

algo/quark/quark-gate.c

@@ -14,7 +14,8 @@ bool register_quark_algo( algo_gate_t* gate )
   gate->scanhash  = (void*)&scanhash_quark;
   gate->hash      = (void*)&quark_hash;
 #endif
-  gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT | VAES_OPT;
+  gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT | VAES_OPT
+                      | NEON_OPT;
   return true;
 };
 

algo/quark/quark-gate.h

@@ -4,7 +4,7 @@
 #include "algo-gate-api.h"
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define QUARK_8WAY 1
 #elif defined(__AVX2__) && defined(__AES__)
   #define QUARK_4WAY 1

algo/quark/quark.c

@@ -7,12 +7,12 @@
 #include <stdint.h>
 #include <string.h>
 #include <stdio.h>
-#include "algo/blake/sph_blake.h"
+#include "algo/blake/blake512-hash.h"
 #include "algo/bmw/sph_bmw.h"
 #include "algo/jh/sph_jh.h"
 #include "algo/keccak/sph_keccak.h"
 #include "algo/skein/sph_skein.h"
-#if defined(__AES__)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
   #include "algo/groestl/aes_ni/hash-groestl.h"
 #else
   #include "algo/groestl/sph_groestl.h"
@@ -21,9 +21,9 @@
 void quark_hash(void *state, const void *input)
 {
    uint32_t hash[16] __attribute__((aligned(64)));
-   sph_blake512_context    ctx_blake;
+   blake512_context        ctx_blake;
    sph_bmw512_context      ctx_bmw;
-#if defined(__AES__)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
    hashState_groestl       ctx_groestl;
 #else
    sph_groestl512_context  ctx_groestl;
@@ -33,17 +33,15 @@ void quark_hash(void *state, const void *input)
    sph_keccak512_context   ctx_keccak;
    uint32_t mask = 8;
 
-   sph_blake512_init( &ctx_blake );
-   sph_blake512( &ctx_blake, input, 80 );
-   sph_blake512_close( &ctx_blake, hash );
-
+   blake512_full( &ctx_blake, hash, input, 80 );
+   
    sph_bmw512_init( &ctx_bmw );
    sph_bmw512( &ctx_bmw, hash, 64 );
    sph_bmw512_close( &ctx_bmw, hash ); 
 
    if ( hash[0] & mask )
    {
-#if defined(__AES__)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
       init_groestl( &ctx_groestl, 64 );
       update_and_final_groestl( &ctx_groestl, (char*)hash,
                                         (const char*)hash, 512 );
@@ -60,7 +58,7 @@ void quark_hash(void *state, const void *input)
       sph_skein512_close( &ctx_skein, hash );
    }
 
-#if defined(__AES__)
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
    init_groestl( &ctx_groestl, 64 );
    update_and_final_groestl( &ctx_groestl, (char*)hash,
                                      (const char*)hash, 512 );
@@ -76,9 +74,7 @@ void quark_hash(void *state, const void *input)
 
    if ( hash[0] & mask )
    {
-      sph_blake512_init( &ctx_blake );
-      sph_blake512( &ctx_blake, hash, 64 );
-      sph_blake512_close( &ctx_blake, hash );
+      blake512_full( &ctx_blake, hash, hash, 64 );
    }
    else
    {

algo/qubit/deep-2way.c

@@ -83,7 +83,7 @@ int scanhash_deep_2way( struct work *work,uint32_t max_nonce,
 
      casti_m256i( endiandata, 0 ) = mm256_bswap_32( casti_m256i( pdata, 0 ) );
      casti_m256i( endiandata, 1 ) = mm256_bswap_32( casti_m256i( pdata, 1 ) );
-     casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
+     casti_v128(  endiandata, 4 ) = v128_bswap32(   casti_v128(  pdata, 4 ) );
 
      uint64_t *edata = (uint64_t*)endiandata;
      intrlv_2x128( (uint64_t*)vdata, edata, edata, 640 );

algo/qubit/qubit-2way.c

@@ -236,7 +236,7 @@ int scanhash_qubit_2way( struct work *work,uint32_t max_nonce,
 
      casti_m256i( endiandata, 0 ) = mm256_bswap_32( casti_m256i( pdata, 0 ) );
      casti_m256i( endiandata, 1 ) = mm256_bswap_32( casti_m256i( pdata, 1 ) );
-     casti_m128i( endiandata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
+     casti_v128(  endiandata, 4 ) = v128_bswap32(   casti_v128(  pdata, 4 ) );
 
      uint64_t *edata = (uint64_t*)endiandata;
      intrlv_2x128( (uint64_t*)vdata, edata, edata, 640 );

algo/qubit/qubit-gate.c

@@ -16,7 +16,8 @@ bool register_qubit_algo( algo_gate_t* gate )
   gate->scanhash  = (void*)&scanhash_qubit;
   gate->hash      = (void*)&qubit_hash;
 #endif
-  gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT | VAES_OPT;
+  gate->optimizations = SSE2_OPT | AES_OPT | AVX2_OPT | AVX512_OPT | VAES_OPT
+                      | NEON_OPT;
   return true;
 };
 

algo/qubit/qubit-gate.h

@@ -5,7 +5,7 @@
 #include <stdint.h>
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define QUBIT_4WAY 1
 #elif defined(__AVX2__) && defined(__AES__)
   #define QUBIT_2WAY 1

algo/qubit/qubit.c

@@ -8,13 +8,9 @@
 #include <stdio.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"
 #include "algo/shavite/sph_shavite.h"
-#ifdef __AES__
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
 #include "algo/echo/aes_ni/hash_api.h"
 #else
 #include "algo/echo/sph_echo.h"
@@ -25,12 +21,8 @@ typedef struct
         hashState_luffa         luffa;
         cubehashParam           cubehash;
         sph_shavite512_context  shavite;
-#if defined(__aarch64__)
-   sph_simd512_context     simd;
-#else
-   hashState_sd            simd;
-#endif
-#ifdef __AES__
+        simd512_context         simd;
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
         hashState_echo          echo;
 #else
         sph_echo512_context echo;
@@ -45,12 +37,7 @@ void init_qubit_ctx()
         init_luffa(&qubit_ctx.luffa,512);
         cubehashInit(&qubit_ctx.cubehash,512,16,32);
         sph_shavite512_init(&qubit_ctx.shavite);
-#if defined(__aarch64__)
-   sph_simd512_init( &qubit_ctx.simd );
-#else
-   init_sd( &qubit_ctx.simd, 512 );
-#endif
-#ifdef __AES__
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
         init_echo(&qubit_ctx.echo, 512);
 #else
         sph_echo512_init(&qubit_ctx.echo);
@@ -81,15 +68,9 @@ void qubit_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_sd( &ctx.simd, (const BitSequence *)hash, 512 );
-    final_sd( &ctx.simd, (BitSequence *)hash );
-#endif
-
-#ifdef __AES__
+        simd512_ctx( &ctx.simd, hash, hash, 64 );
+        
+#if defined(__AES__) || defined(__ARM_FEATURE_AES)
         update_final_echo( &ctx.echo, (BitSequence *) hash,
                      (const BitSequence *) hash, 512 );
 #else

algo/ripemd/lbry-4way.c

@@ -104,14 +104,14 @@ int scanhash_lbry_16way( struct work *work, uint32_t max_nonce,
    int thr_id = mythr->id;  // thr_id arg is deprecated
 
    // we need bigendian data...
-   casti_m128i( edata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
-   casti_m128i( edata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
-   casti_m128i( edata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
-   casti_m128i( edata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
-   casti_m128i( edata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
-   casti_m128i( edata, 5 ) = mm128_bswap_32( casti_m128i( pdata, 5 ) );
-   casti_m128i( edata, 6 ) = mm128_bswap_32( casti_m128i( pdata, 6 ) );
-   casti_m128i( edata, 7 ) = mm128_bswap_32( casti_m128i( pdata, 7 ) );
+   casti_v128u32( edata, 0 ) = v128_bswap32( casti_v128u32( pdata, 0 ) );
+   casti_v128u32( edata, 1 ) = v128_bswap32( casti_v128u32( pdata, 1 ) );
+   casti_v128u32( edata, 2 ) = v128_bswap32( casti_v128u32( pdata, 2 ) );
+   casti_v128u32( edata, 3 ) = v128_bswap32( casti_v128u32( pdata, 3 ) );
+   casti_v128u32( edata, 4 ) = v128_bswap32( casti_v128u32( pdata, 4 ) );
+   casti_v128u32( edata, 5 ) = v128_bswap32( casti_v128u32( pdata, 5 ) );
+   casti_v128u32( edata, 6 ) = v128_bswap32( casti_v128u32( pdata, 6 ) );
+   casti_v128u32( edata, 7 ) = v128_bswap32( casti_v128u32( pdata, 7 ) );
    intrlv_16x32( vdata, edata, edata, edata, edata, edata, edata, edata,
         edata, edata, edata, edata, edata, edata, edata, edata, edata, 1024 );
 
@@ -224,14 +224,14 @@ int scanhash_lbry_8way( struct work *work, uint32_t max_nonce,
    int thr_id = mythr->id;  // thr_id arg is deprecated
 
    // we need bigendian data...
-   casti_m128i( edata, 0 ) = mm128_bswap_32( casti_m128i( pdata, 0 ) );
-   casti_m128i( edata, 1 ) = mm128_bswap_32( casti_m128i( pdata, 1 ) );
-   casti_m128i( edata, 2 ) = mm128_bswap_32( casti_m128i( pdata, 2 ) );
-   casti_m128i( edata, 3 ) = mm128_bswap_32( casti_m128i( pdata, 3 ) );
-   casti_m128i( edata, 4 ) = mm128_bswap_32( casti_m128i( pdata, 4 ) );
-   casti_m128i( edata, 5 ) = mm128_bswap_32( casti_m128i( pdata, 5 ) );
-   casti_m128i( edata, 6 ) = mm128_bswap_32( casti_m128i( pdata, 6 ) );
-   casti_m128i( edata, 7 ) = mm128_bswap_32( casti_m128i( pdata, 7 ) );
+   casti_v128u32( edata, 0 ) = v128_bswap32( casti_v128u32( pdata, 0 ) );
+   casti_v128u32( edata, 1 ) = v128_bswap32( casti_v128u32( pdata, 1 ) );
+   casti_v128u32( edata, 2 ) = v128_bswap32( casti_v128u32( pdata, 2 ) );
+   casti_v128u32( edata, 3 ) = v128_bswap32( casti_v128u32( pdata, 3 ) );
+   casti_v128u32( edata, 4 ) = v128_bswap32( casti_v128u32( pdata, 4 ) );
+   casti_v128u32( edata, 5 ) = v128_bswap32( casti_v128u32( pdata, 5 ) );
+   casti_v128u32( edata, 6 ) = v128_bswap32( casti_v128u32( pdata, 6 ) );
+   casti_v128u32( edata, 7 ) = v128_bswap32( casti_v128u32( pdata, 7 ) );
    intrlv_8x32( vdata, edata, edata, edata, edata,
                        edata, edata, edata, edata, 1024 );
 

algo/ripemd/lbry-gate.c

@@ -51,7 +51,6 @@ int lbry_get_work_data_size() { return LBRY_WORK_DATA_SIZE; }
 
 bool register_lbry_algo( algo_gate_t* gate )
 {
-//  gate->optimizations = AVX2_OPT | AVX512_OPT | SHA_OPT;
 #if defined (LBRY_16WAY)
   gate->scanhash              = (void*)&scanhash_lbry_16way;
   gate->hash                  = (void*)&lbry_16way_hash;
@@ -67,7 +66,7 @@ bool register_lbry_algo( algo_gate_t* gate )
 #else 
   gate->scanhash              = (void*)&scanhash_lbry;
   gate->hash                  = (void*)&lbry_hash;
-  gate->optimizations = AVX2_OPT | AVX512_OPT | SHA_OPT;
+  gate->optimizations = AVX2_OPT | AVX512_OPT | SHA256_OPT;
 #endif
   gate->build_stratum_request = (void*)&lbry_le_build_stratum_request;
   gate->build_extraheader     = (void*)&lbry_build_extraheader;

algo/ripemd/lbry-gate.h

@@ -5,7 +5,7 @@
 #include <stdint.h>
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define LBRY_16WAY 1
 #elif defined(__AVX2__)
   #define LBRY_8WAY 1

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

@@ -35,20 +35,20 @@ static const uint32_t IV[5] =
    _mm_xor_si128( _mm_and_si128( _mm_xor_si128( y, z ), x ), z )
 
 #define F3(x, y, z) \
-   _mm_xor_si128( _mm_or_si128( x, v128_not( y ) ), z )
+   _mm_xor_si128( v128_ornot( y, x ), z )
 
 #define F4(x, y, z) \
    _mm_xor_si128( _mm_and_si128( _mm_xor_si128( x, y ), z ), y )
 
 #define F5(x, y, z) \
-   _mm_xor_si128( x, _mm_or_si128( y, v128_not( z ) ) )
+   _mm_xor_si128( x, v128_ornot( z, y ) )
 
 #define RR(a, b, c, d, e, f, s, r, k) \
 do{ \
-   a = _mm_add_epi32( mm128_rol_32( _mm_add_epi32( _mm_add_epi32( \
+   a = _mm_add_epi32( v128_rol32( _mm_add_epi32( _mm_add_epi32( \
                 _mm_add_epi32( a, f( b ,c, d ) ), r ), \
                                  _mm_set1_epi64x( k ) ), s ), e ); \
-   c = mm128_rol_32( c, 10 );\
+   c = v128_rol32( c, 10 );\
 } while (0)
 
 #define ROUND1(a, b, c, d, e, f, s, r, k)  \
@@ -319,7 +319,7 @@ void ripemd160_4way_close( ripemd160_4way_context  *sc, void *dst )
     sc->buf[ (pad>>2) + 1 ] = _mm_set1_epi32( high );
     ripemd160_4way_round( sc );
     for (u = 0; u < 5; u ++)
-        casti_m128i( dst, u ) = sc->val[u];
+        casti_v128u32( dst, u ) = sc->val[u];
 }
 
 #endif
@@ -335,13 +335,13 @@ void ripemd160_4way_close( ripemd160_4way_context  *sc, void *dst )
    _mm256_xor_si256( _mm256_and_si256( _mm256_xor_si256( y, z ), x ), z )
 
 #define F8W_3(x, y, z) \
-   _mm256_xor_si256( _mm256_or_si256( x, mm256_not( y ) ), z )
+   _mm256_xor_si256( mm256_ornot( y, x ), z )
 
 #define F8W_4(x, y, z) \
    _mm256_xor_si256( _mm256_and_si256( _mm256_xor_si256( x, y ), z ), y )
 
 #define F8W_5(x, y, z) \
-   _mm256_xor_si256( x, _mm256_or_si256( y, mm256_not( z ) ) )
+   _mm256_xor_si256( x, mm256_ornot( z, y ) )
 
 #define RR_8W(a, b, c, d, e, f, s, r, k) \
 do{ \
@@ -625,7 +625,7 @@ void ripemd160_8way_close( ripemd160_8way_context  *sc, void *dst )
 
 #endif // __AVX2__
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 //  RIPEMD-160 16 way
 

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

@@ -33,7 +33,7 @@ void ripemd160_8way_update( ripemd160_8way_context *sc, const void *data,
                             size_t len );
 void ripemd160_8way_close( ripemd160_8way_context *sc, void *dst );
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct
 {

algo/scrypt/scrypt-core-4way.c

@@ -745,7 +745,7 @@ do{ \
    SALSA_2ROUNDS; SALSA_2ROUNDS; SALSA_2ROUNDS; SALSA_2ROUNDS;
 
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // Tested OK but very slow
 // 16 way parallel, requires 16x32 interleaving
@@ -2074,7 +2074,7 @@ void scrypt_core_4way( v128_t *X, v128_t *V, const uint32_t N )
          v128_ovly v;    
          for ( int l = 0; l < 4; l++ )
             v.u32[l] = ( *(vptr[l] +i ) ) .u32[l];
-         X[i] = v128_xor( X[i], v.m128 );
+         X[i] = v128_xor( X[i], v.v128 );
       }
 
       xor_salsa8_4way( &X[ 0], &X[16] );
@@ -2211,10 +2211,10 @@ static void salsa8_simd128( uint32_t *b, const uint32_t * const c)
    // X2 is shuffled left 2 (swap_64)    { xd, x8, x7, x2 }
    // X3 is shuffled left 3 (ror_1x32)   { xc, xb, x6, x1 }
 
-   y[0].m128 = X0;
-   y[1].m128 = X1;
-   y[2].m128 = X2;
-   y[3].m128 = X3;
+   y[0].v128 = X0;
+   y[1].v128 = X1;
+   y[2].v128 = X2;
+   y[3].v128 = X3;
 
    z[0].u32[0] = y[0].u32[0];
    z[0].u32[3] = y[1].u32[0];
@@ -2236,10 +2236,10 @@ static void salsa8_simd128( uint32_t *b, const uint32_t * const c)
    z[3].u32[1] = y[2].u32[3];
    z[3].u32[0] = y[3].u32[3];
 
-   B[0] = v128_add32( B[0], z[0].m128 );
-   B[1] = v128_add32( B[1], z[1].m128 );
-   B[2] = v128_add32( B[2], z[2].m128 );
-   B[3] = v128_add32( B[3], z[3].m128 );
+   B[0] = v128_add32( B[0], z[0].v128 );
+   B[1] = v128_add32( B[1], z[1].v128 );
+   B[2] = v128_add32( B[2], z[2].v128 );
+   B[3] = v128_add32( B[3], z[3].v128 );
 
 #endif
 
@@ -2404,14 +2404,14 @@ static inline void salsa_simd128_unshuffle_2buf( uint32_t* xa, uint32_t* xb )
 /*
    v128_ovly ya[4], za[4], yb[4], zb[4];
 
-   ya[0].m128 = XA[0];
-   yb[0].m128 = XB[0];
-   ya[1].m128 = XA[1];
-   yb[1].m128 = XB[1];
-   ya[2].m128 = XA[2];
-   yb[2].m128 = XB[2];
-   ya[3].m128 = XA[3];
-   yb[3].m128 = XB[3];
+   ya[0].v128 = XA[0];
+   yb[0].v128 = XB[0];
+   ya[1].v128 = XA[1];
+   yb[1].v128 = XB[1];
+   ya[2].v128 = XA[2];
+   yb[2].v128 = XB[2];
+   ya[3].v128 = XA[3];
+   yb[3].v128 = XB[3];
 
    za[0].u32[0] = ya[0].u32[0];
    zb[0].u32[0] = yb[0].u32[0];
@@ -2449,14 +2449,14 @@ static inline void salsa_simd128_unshuffle_2buf( uint32_t* xa, uint32_t* xb )
    za[3].u32[3] = ya[0].u32[3];
    zb[3].u32[3] = yb[0].u32[3];
 
-   XA[0] = za[0].m128;
-   XB[0] = zb[0].m128;
-   XA[1] = za[1].m128;
-   XB[1] = zb[1].m128;
-   XA[2] = za[2].m128;
-   XB[2] = zb[2].m128;
-   XA[3] = za[3].m128;
-   XB[3] = zb[3].m128;
+   XA[0] = za[0].v128;
+   XB[0] = zb[0].v128;
+   XA[1] = za[1].v128;
+   XB[1] = zb[1].v128;
+   XA[2] = za[2].v128;
+   XB[2] = zb[2].v128;
+   XA[3] = za[3].v128;
+   XB[3] = zb[3].v128;
 */
 }
 
@@ -2487,7 +2487,7 @@ static void salsa8_simd128_2buf( uint32_t * const ba, uint32_t * const bb,
    XA3 = BA[3] = v128_xor( BA[3], CA[3] );
    XB3 = BB[3] = v128_xor( BB[3], CB[3] );
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
              
    SALSA_8ROUNDS_SIMD128_2BUF;
 
@@ -2770,18 +2770,18 @@ static inline void salsa_simd128_unshuffle_3buf( uint32_t* xa, uint32_t* xb,
 /*  
    v128_ovly ya[4], za[4], yb[4], zb[4], yc[4], zc[4];
 
-   ya[0].m128 = XA[0];
-   yb[0].m128 = XB[0];
-   yc[0].m128 = XC[0];
-   ya[1].m128 = XA[1];
-   yb[1].m128 = XB[1];
-   yc[1].m128 = XC[1];
-   ya[2].m128 = XA[2];
-   yb[2].m128 = XB[2];
-   yc[2].m128 = XC[2];
-   ya[3].m128 = XA[3];
-   yb[3].m128 = XB[3];
-   yc[3].m128 = XC[3];
+   ya[0].v128 = XA[0];
+   yb[0].v128 = XB[0];
+   yc[0].v128 = XC[0];
+   ya[1].v128 = XA[1];
+   yb[1].v128 = XB[1];
+   yc[1].v128 = XC[1];
+   ya[2].v128 = XA[2];
+   yb[2].v128 = XB[2];
+   yc[2].v128 = XC[2];
+   ya[3].v128 = XA[3];
+   yb[3].v128 = XB[3];
+   yc[3].v128 = XC[3];
 
    za[0].u32[0] = ya[0].u32[0];
    zb[0].u32[0] = yb[0].u32[0];
@@ -2835,18 +2835,18 @@ static inline void salsa_simd128_unshuffle_3buf( uint32_t* xa, uint32_t* xb,
    zb[3].u32[3] = yb[0].u32[3];
    zc[3].u32[3] = yc[0].u32[3];
 
-   XA[0] = za[0].m128;
-   XB[0] = zb[0].m128;
-   XC[0] = zc[0].m128;
-   XA[1] = za[1].m128;
-   XB[1] = zb[1].m128;
-   XC[1] = zc[1].m128;
-   XA[2] = za[2].m128;
-   XB[2] = zb[2].m128;
-   XC[2] = zc[2].m128;
-   XA[3] = za[3].m128;
-   XB[3] = zb[3].m128;
-   XC[3] = zc[3].m128;
+   XA[0] = za[0].v128;
+   XB[0] = zb[0].v128;
+   XC[0] = zc[0].v128;
+   XA[1] = za[1].v128;
+   XB[1] = zb[1].v128;
+   XC[1] = zc[1].v128;
+   XA[2] = za[2].v128;
+   XB[2] = zb[2].v128;
+   XC[2] = zc[2].v128;
+   XA[3] = za[3].v128;
+   XB[3] = zb[3].v128;
+   XC[3] = zc[3].v128;
 */
 }   
 
@@ -2886,7 +2886,7 @@ static void salsa8_simd128_3buf( uint32_t *ba, uint32_t *bb, uint32_t *bc,
    XB3 = BB[3] = v128_xor( BB[3], CB[3] );
    XC3 = BC[3] = v128_xor( BC[3], CC[3] );
       
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
    
    SALSA_8ROUNDS_SIMD128_3BUF;
 
@@ -3049,7 +3049,7 @@ static void xor_salsa8(uint32_t * const B, const uint32_t * const C)
             xf = (B[15] ^= C[15]);
 
    
-   #define ROL32( a, c )    ror32( a, c )
+   #define ROL32( a, c )    rol32( a, c )
    #define ADD32( a, b )    ( (a)+(b) )
    #define XOR( a, b )      ( (a)^(b) )
 

algo/scrypt/scrypt-core-4way.h

@@ -5,7 +5,7 @@
 #include <stdlib.h>
 #include <stdint.h>
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 void scrypt_core_16way( __m512i *X, __m512i *V, const uint32_t N );
 

algo/scrypt/scrypt.c

@@ -35,7 +35,7 @@
 //#include <mm_malloc.h>
 #include "malloc-huge.h"
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
   #define SCRYPT_THROUGHPUT 16
 #elif defined(__SHA__) || defined(__ARM_FEATURE_SHA2)
   #define SCRYPT_THROUGHPUT 2
@@ -592,7 +592,7 @@ static inline void PBKDF2_SHA256_128_32_8way( uint32_t *tstate,
 
 #endif /* HAVE_SHA256_8WAY */
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 static inline void sha256_16way_init_state( void *state )
 {
@@ -1481,7 +1481,7 @@ bool scrypt_miner_thread_init( int thr_id )
 bool register_scrypt_algo( algo_gate_t* gate )
 {
 #if defined(__SHA__) || defined(__ARM_FEATURE_SHA2)
-   gate->optimizations = SSE2_OPT | SHA_OPT | NEON_OPT;
+   gate->optimizations = SSE2_OPT | SHA256_OPT | NEON_OPT;
 #else
    gate->optimizations = SSE2_OPT | SSE42_OPT | AVX_OPT | AVX2_OPT | AVX512_OPT | NEON_OPT;
 #endif
@@ -1494,7 +1494,7 @@ bool register_scrypt_algo( algo_gate_t* gate )
 // scrypt_throughput defined at compile time and used to replace
 // MAX_WAYS to reduce memory usage.
    
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 //   scrypt_throughput = 16;
    if ( opt_param_n > 0x4000 )
       scratchbuf_size = opt_param_n * 3 * 128;  // 3 buf

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

@@ -74,8 +74,8 @@ hmac_sha256_4way_init( hmac_sha256_4way_context *ctx, const void *_K,
 	memset( pad, 0x36, 64*4 );
 
    for ( i = 0; i < Klen; i++ )
-		casti_m128i( pad, i ) = _mm_xor_si128( casti_m128i( pad, i ),
-                                             casti_m128i( K, i ) );
+		casti_v128u32( pad, i ) = _mm_xor_si128( casti_v128u32( pad, i ),
+                                               casti_v128u32( K, i ) );
 
    sha256_4way_update( &ctx->ictx, pad, 64 );
 
@@ -83,8 +83,8 @@ hmac_sha256_4way_init( hmac_sha256_4way_context *ctx, const void *_K,
 	sha256_4way_init( &ctx->octx );
 	memset( pad, 0x5c, 64*4 );
 	for ( i = 0; i < Klen/4; i++ )
-		casti_m128i( pad, i ) = _mm_xor_si128( casti_m128i( pad, i ),
-                                             casti_m128i( K, i ) );
+		casti_v128u32( pad, i ) = _mm_xor_si128( casti_v128u32( pad, i ),
+                                               casti_v128u32( K, i ) );
 	sha256_4way_update( &ctx->octx, pad, 64 );
 }
 
@@ -158,8 +158,8 @@ pbkdf2_sha256_4way( uint8_t *buf, size_t dkLen,
 
 			/* ... xor U_j ... */
 			for ( k = 0; k < 8; k++ )
-				casti_m128i( T, k ) = _mm_xor_si128( casti_m128i( T, k ),
-                                                 casti_m128i( U, k ) );
+				casti_v128u32( T, k ) = _mm_xor_si128( casti_v128u32( T, k ),
+                                                   casti_v128u32( U, k ) );
 		}
 
 		/* Copy as many bytes as necessary into buf. */
@@ -306,7 +306,7 @@ pbkdf2_sha256_8way( uint8_t *buf, size_t dkLen, const uint8_t *passwd,
    }
 }
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // HMAC 16-way AVX512
 

algo/sha/hmac-sha256-hash-4way.h

@@ -84,7 +84,7 @@ void pbkdf2_sha256_8way( uint8_t *, size_t, const uint8_t *, size_t,
 
 #endif  // AVX2
        
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 typedef struct _hmac_sha256_16way_context
 {

algo/sha/sha1-hash.c

@@ -205,7 +205,7 @@ void sha1_x86_sha_transform_be( uint32_t *state_out, const void *input,
 
 #endif
 
-#if defined(__aarch64__) && defined(__ARM_FEATURE_SHA2)
+#if defined(__ARM_NEON) && defined(__ARM_FEATURE_SHA2)
 
 #define sha1_neon_rounds( state_out, data, state_in ) \
 { \

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

@@ -580,7 +580,7 @@ void sha256_4x32_full( void *dst, const void *data, size_t len )
 // to avoid recalculating it as Y^Z. This optimization is not applicable
 // when MAJ is optimized with ternary logic.
 
-#if defined(__AVX512VL__)
+#if defined(VL256)
 
 #define CHx(X, Y, Z)    _mm256_ternarylogic_epi32( X, Y, Z, 0xca )
 
@@ -788,7 +788,7 @@ void sha256_8way_prehash_3rounds( __m256i *state_mid, __m256i *X,
    G = _mm256_load_si256( state_in + 6 );
    H = _mm256_load_si256( state_in + 7 );
 
-#if !defined(__AVX512VL__)
+#if !defined(VL256)
    __m256i X_xor_Y, Y_xor_Z = _mm256_xor_si256( B, C );
 #endif
 
@@ -830,7 +830,7 @@ void sha256_8way_final_rounds( __m256i *state_out, const __m256i *data,
    G = _mm256_load_si256( state_mid + 6 );
    H = _mm256_load_si256( state_mid + 7 );
 
-#if !defined(__AVX512VL__)
+#if !defined(VL256)
    __m256i X_xor_Y, Y_xor_Z = _mm256_xor_si256( F, G );
 #endif
 
@@ -936,7 +936,7 @@ int sha256_8way_transform_le_short( __m256i *state_out, const __m256i *data,
    const __m256i IV7 = H;
    const __m256i IV6 = G;
 
-#if !defined(__AVX512VL__)
+#if !defined(VL256)
    __m256i X_xor_Y, Y_xor_Z = _mm256_xor_si256( B, C );
 #endif
 
@@ -981,7 +981,7 @@ int sha256_8way_transform_le_short( __m256i *state_out, const __m256i *data,
    W[11] = SHA256_8WAY_MEXP( W[ 9], W[ 4], W[12], W[11] );
    W[12] = SHA256_8WAY_MEXP( W[10], W[ 5], W[13], W[12] );
 
-#if !defined(__AVX512VL__)
+#if !defined(VL256)
    Y_xor_Z = _mm256_xor_si256( B, C );
 #endif
 
@@ -1172,7 +1172,7 @@ void sha256_8way_full( void *dst, const void *data, size_t len )
    sha256_8way_close( &ctx, dst );
 }
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
 // SHA-256 16 way
 

algo/sha/sha256-hash.c

@@ -1,6 +1,6 @@
 #include "sha256-hash.h"
 
-#if ( defined(__x86_64__) && defined(__SHA__) ) || defined(__ARM_NEON) && defined(__ARM_FEATURE_SHA2)
+#if ( defined(__x86_64__) && defined(__SHA__) ) || ( defined(__ARM_NEON) && defined(__ARM_FEATURE_SHA2) )
 
 static const uint32_t SHA256_IV[8] =
 {
@@ -10,6 +10,28 @@ static const uint32_t SHA256_IV[8] =
 
 #if defined(__x86_64__) && defined(__SHA__)
 
+
+/* common code used for rounds 12 through 51  */
+
+#define sha256_generic_qround( s0, s1, m, a, b, c ) \
+    TMP = _mm_alignr_epi8( a, c, 4 ); \
+    s1 = _mm_sha256rnds2_epu32( s1, s0, m ); \
+    b  = _mm_add_epi32( b, TMP ); \
+    b  = _mm_sha256msg2_epu32( b, a ); \
+    m  = _mm_shuffle_epi32( m, 0x0e ); \
+    s0 = _mm_sha256rnds2_epu32( s0, s1, m ); \
+    c  = _mm_sha256msg1_epu32( c, a );
+
+// r12-15      
+// sha256_generic_qround( s0, s1, m, t3, t0, t2 )
+// r16-19
+// sha256_generic_qround( s0, s1, m, t0, t1, t3 )
+// r20-23
+// sha256_generic_qround( s0, s1, m, t1, t2, t0 ) 
+// r24-27
+// sha256_generic_qround( s0, s1, m, t2, t3, t1 ) ... 
+
+
 #define sha256_opt_rounds( state_out, input, state_in ) \
 { \
     __m128i STATE0, STATE1; \
@@ -189,7 +211,7 @@ static const uint32_t SHA256_IV[8] =
     _mm_store_si128( (__m128i*) &state_out[4], STATE1 ); \
 }
 
-void sha256_opt_transform_le( uint32_t *state_out, const void *input,
+void sha256_x86_sha_transform_le( uint32_t *state_out, const void *input,
                               const uint32_t *state_in )
 {
 #define load_msg( m, i ) casti_v128( m, i )
@@ -197,7 +219,7 @@ void sha256_opt_transform_le( uint32_t *state_out, const void *input,
 #undef load_msg
 }
 
-void sha256_opt_transform_be( uint32_t *state_out, const void *input,
+void sha256_x86_sha_transform_be( uint32_t *state_out, const void *input,
                               const uint32_t *state_in )
 {
 #define load_msg( m, i ) v128_bswap32( casti_v128( m, i ) )
@@ -517,7 +539,7 @@ void sha256_opt_transform_be( uint32_t *state_out, const void *input,
     _mm_store_si128( (__m128i*) &out_Y[4], STATE1_Y ); \
 }
 
-void sha256_ni2x_transform_le( uint32_t *out_X, uint32_t*out_Y,
+void sha256_x86_x2sha_transform_le( uint32_t *out_X, uint32_t*out_Y,
                                  const void *msg_X, const void *msg_Y,
                                  const uint32_t *in_X, const uint32_t *in_Y )
 {
@@ -526,7 +548,7 @@ void sha256_ni2x_transform_le( uint32_t *out_X, uint32_t*out_Y,
 #undef load_msg
 }
 
-void sha256_ni2x_transform_be( uint32_t *out_X, uint32_t*out_Y,
+void sha256_x86_x2sha_transform_be( uint32_t *out_X, uint32_t*out_Y,
                               const void *msg_X, const void *msg_Y,
                               const uint32_t *in_X, const uint32_t *in_Y )
 {
@@ -541,14 +563,14 @@ void sha256_ni2x_transform_be( uint32_t *out_X, uint32_t*out_Y,
 // The goal is to avoid any redundant processing in final. Prehash is almost
 // 4 rounds total, only missing the final addition of the nonce.
 // Nonce must be set to zero for prehash.
-void sha256_ni_prehash_3rounds( uint32_t *ostate, const void *msg,
+void sha256_x86_sha_prehash_3rounds( uint32_t *ostate, const void *msg,
                                 uint32_t *sstate, const uint32_t *istate )
 {
    __m128i STATE0, STATE1, MSG, TMP;
 
    // Load initial values
-   TMP    = casti_m128i( istate, 0 );
-   STATE1 = casti_m128i( istate, 1 );
+   TMP    = casti_v128u32( istate, 0 );
+   STATE1 = casti_v128u32( istate, 1 );
 
    TMP    = _mm_shuffle_epi32( TMP, 0xB1 );       // CDAB
    STATE1 = _mm_shuffle_epi32( STATE1, 0x1B );    // EFGH
@@ -556,20 +578,20 @@ void sha256_ni_prehash_3rounds( uint32_t *ostate, const void *msg,
    STATE1 = _mm_blend_epi16( STATE1, TMP, 0xF0 ); // CDGH
 
    // Save current hash
-   casti_m128i( sstate, 0 ) = STATE0;
-   casti_m128i( sstate, 1 ) = STATE1;
+   casti_v128u32( sstate, 0 ) = STATE0;
+   casti_v128u32( sstate, 1 ) = STATE1;
 
    // Rounds 0 to 3
-   MSG = casti_m128i( msg, 0 );
+   MSG = casti_v128u32( msg, 0 );
    TMP = _mm_set_epi64x( 0xE9B5DBA5B5C0FBCFULL, 0x71374491428A2F98ULL );
    MSG = _mm_add_epi32( MSG, TMP );
    STATE1 = _mm_sha256rnds2_epu32( STATE1, STATE0, MSG );
    MSG = _mm_shuffle_epi32( MSG, 0x0E );
-   casti_m128i( ostate, 0 ) = _mm_sha256rnds2_epu32( STATE0, STATE1, MSG );
-   casti_m128i( ostate, 1 ) = STATE1;
+   casti_v128u32( ostate, 0 ) = _mm_sha256rnds2_epu32( STATE0, STATE1, MSG );
+   casti_v128u32( ostate, 1 ) = STATE1;
 }
 
-void sha256_ni2x_final_rounds( uint32_t *out_X, uint32_t *out_Y,
+void sha256_x86_x2sha_final_rounds( uint32_t *out_X, uint32_t *out_Y,
                  const void *msg_X, const void *msg_Y,
                  const uint32_t *state_mid_X, const uint32_t *state_mid_Y,
                  const uint32_t *state_save_X, const uint32_t *state_save_Y )
@@ -579,22 +601,22 @@ void sha256_ni2x_final_rounds( uint32_t *out_X, uint32_t *out_Y,
     __m128i TMSG0_X, TMSG1_X, TMSG2_X, TMSG3_X;
     __m128i TMSG0_Y, TMSG1_Y, TMSG2_Y, TMSG3_Y;
 
-    STATE0_X = casti_m128i( state_mid_X, 0 );
-    STATE1_X = casti_m128i( state_mid_X, 1 );
-    STATE0_Y = casti_m128i( state_mid_Y, 0 );
-    STATE1_Y = casti_m128i( state_mid_Y, 1 );
+    STATE0_X = casti_v128u32( state_mid_X, 0 );
+    STATE1_X = casti_v128u32( state_mid_X, 1 );
+    STATE0_Y = casti_v128u32( state_mid_Y, 0 );
+    STATE1_Y = casti_v128u32( state_mid_Y, 1 );
 
     // Add the nonces (msg[0] lane 3) to A & E (STATE0 lanes 1 & 3)
-    TMSG0_X = casti_m128i( msg_X, 0 );
-    TMSG0_Y = casti_m128i( msg_Y, 0 );
-    TMP_X = mm128_xim_32( TMSG0_X, TMSG0_X, 0xd5 );
-    TMP_Y = mm128_xim_32( TMSG0_Y, TMSG0_Y, 0xd5 );
+    TMSG0_X = casti_v128u32( msg_X, 0 );
+    TMSG0_Y = casti_v128u32( msg_Y, 0 );
+    TMP_X = v128_xim32( TMSG0_X, TMSG0_X, 0xd5 );
+    TMP_Y = v128_xim32( TMSG0_Y, TMSG0_Y, 0xd5 );
     STATE0_X = _mm_add_epi32( STATE0_X, TMP_X );
     STATE0_Y = _mm_add_epi32( STATE0_Y, TMP_Y );
 
     // Rounds 4 to 7
-    TMSG1_X = casti_m128i( msg_X, 1 );
-    TMSG1_Y = casti_m128i( msg_Y, 1 );
+    TMSG1_X = casti_v128u32( msg_X, 1 );
+    TMSG1_Y = casti_v128u32( msg_Y, 1 );
     TMP_X = _mm_set_epi64x( 0xAB1C5ED5923F82A4ULL, 0x59F111F13956C25BULL );
     MSG_X = _mm_add_epi32( TMSG1_X, TMP_X );
     MSG_Y = _mm_add_epi32( TMSG1_Y, TMP_X );
@@ -616,8 +638,8 @@ void sha256_ni2x_final_rounds( uint32_t *out_X, uint32_t *out_Y,
     STATE0_Y = _mm_sha256rnds2_epu32( STATE0_Y, STATE1_Y, MSG_X );
 
     // Rounds 12 to 15
-    TMSG3_X = casti_m128i( msg_X, 3 );
-    TMSG3_Y = casti_m128i( msg_Y, 3 );
+    TMSG3_X = casti_v128u32( msg_X, 3 );
+    TMSG3_Y = casti_v128u32( msg_Y, 3 );
     TMP_X = _mm_set_epi64x( 0xC19BF1749BDC06A7ULL, 0x80DEB1FE72BE5D74ULL );
     MSG_X = _mm_add_epi32( TMSG3_X, TMP_X );
     MSG_Y = _mm_add_epi32( TMSG3_Y, TMP_X );
@@ -845,20 +867,20 @@ void sha256_ni2x_final_rounds( uint32_t *out_X, uint32_t *out_Y,
     STATE0_Y = _mm_sha256rnds2_epu32( STATE0_Y, STATE1_Y, MSG_Y );
 
     // Add saved state to new state
-    STATE0_X = _mm_add_epi32( STATE0_X, casti_m128i( state_save_X, 0 ) );
-    STATE1_X = _mm_add_epi32( STATE1_X, casti_m128i( state_save_X, 1 ) );
-    STATE0_Y = _mm_add_epi32( STATE0_Y, casti_m128i( state_save_Y, 0 ) );
-    STATE1_Y = _mm_add_epi32( STATE1_Y, casti_m128i( state_save_Y, 1 ) );
+    STATE0_X = _mm_add_epi32( STATE0_X, casti_v128u32( state_save_X, 0 ) );
+    STATE1_X = _mm_add_epi32( STATE1_X, casti_v128u32( state_save_X, 1 ) );
+    STATE0_Y = _mm_add_epi32( STATE0_Y, casti_v128u32( state_save_Y, 0 ) );
+    STATE1_Y = _mm_add_epi32( STATE1_Y, casti_v128u32( state_save_Y, 1 ) );
 
     // Unshuffle & save state    
     TMP_X = _mm_shuffle_epi32( STATE0_X, 0x1B );                        // FEBA
     TMP_Y = _mm_shuffle_epi32( STATE0_Y, 0x1B );
     STATE1_X = _mm_shuffle_epi32( STATE1_X, 0xB1 );                     // DCHG
     STATE1_Y = _mm_shuffle_epi32( STATE1_Y, 0xB1 );
-    casti_m128i( out_X, 0 ) = _mm_blend_epi16( TMP_X, STATE1_X, 0xF0 ); // DCBA
-    casti_m128i( out_Y, 0 ) = _mm_blend_epi16( TMP_Y, STATE1_Y, 0xF0 );
-    casti_m128i( out_X, 1 ) = _mm_alignr_epi8( STATE1_X, TMP_X, 8 );    // ABEF
-    casti_m128i( out_Y, 1 ) = _mm_alignr_epi8( STATE1_Y, TMP_Y, 8 );
+    casti_v128u32( out_X, 0 ) = _mm_blend_epi16( TMP_X, STATE1_X, 0xF0 ); // DCBA
+    casti_v128u32( out_Y, 0 ) = _mm_blend_epi16( TMP_Y, STATE1_Y, 0xF0 );
+    casti_v128u32( out_X, 1 ) = _mm_alignr_epi8( STATE1_X, TMP_X, 8 );    // ABEF
+    casti_v128u32( out_Y, 1 ) = _mm_alignr_epi8( STATE1_Y, TMP_Y, 8 );
 }
 
 #endif     // SHA
@@ -887,14 +909,14 @@ static const uint32_t K256[64] =
 
 #define sha256_neon_rounds( state_out, input, state_in ) \
 { \
-    uint32x4_t STATE0, STATE1, ABEF_SAVE, CDGH_SAVE; \
+    uint32x4_t STATE0, STATE1, ABCD_SAVE, EFGH_SAVE; \
     uint32x4_t MSG0, MSG1, MSG2, MSG3; \
     uint32x4_t TMP0, TMP1, TMP2; \
 \
     STATE0 = vld1q_u32( state_in   ); \
     STATE1 = vld1q_u32( state_in+4 ); \
-    ABEF_SAVE = STATE0; \
-    CDGH_SAVE = STATE1; \
+    ABCD_SAVE = STATE0; \
+    EFGH_SAVE = STATE1; \
 \
     MSG0 = load_msg( input, 0 ); \
     MSG1 = load_msg( input, 1 ); \
@@ -1004,8 +1026,8 @@ static const uint32_t K256[64] =
     TMP2 = STATE0; \
     STATE0 = vsha256hq_u32( STATE0, STATE1, TMP1 ); \
     STATE1 = vsha256h2q_u32( STATE1, TMP2, TMP1 ); \
-    STATE0 = vaddq_u32( STATE0, ABEF_SAVE ); \
-    STATE1 = vaddq_u32( STATE1, CDGH_SAVE ); \
+    STATE0 = vaddq_u32( STATE0, ABCD_SAVE ); \
+    STATE1 = vaddq_u32( STATE1, EFGH_SAVE ); \
     vst1q_u32( state_out  , STATE0 ); \
     vst1q_u32( state_out+4, STATE1 ); \
 }
@@ -1029,8 +1051,8 @@ void sha256_neon_sha_transform_le( uint32_t *state_out, const void *input,
 #define sha256_neon_x2sha_rounds( state_out_X, state_out_Y, input_X, \
                               input_Y, state_in_X, state_in_Y ) \
 { \
-    uint32x4_t STATE0_X, STATE1_X, ABEF_SAVE_X, CDGH_SAVE_X; \
-    uint32x4_t STATE0_Y, STATE1_Y, ABEF_SAVE_Y, CDGH_SAVE_Y; \
+    uint32x4_t STATE0_X, STATE1_X, ABCD_SAVE_X, EFGH_SAVE_X; \
+    uint32x4_t STATE0_Y, STATE1_Y, ABCD_SAVE_Y, EFGH_SAVE_Y; \
     uint32x4_t MSG0_X, MSG1_X, MSG2_X, MSG3_X; \
     uint32x4_t MSG0_Y, MSG1_Y, MSG2_Y, MSG3_Y; \
     uint32x4_t TMP0_X, TMP1_X, TMP2_X; \
@@ -1040,10 +1062,10 @@ void sha256_neon_sha_transform_le( uint32_t *state_out, const void *input,
     STATE0_Y = vld1q_u32( state_in_Y   ); \
     STATE1_X = vld1q_u32( state_in_X+4 ); \
     STATE1_Y = vld1q_u32( state_in_Y+4 ); \
-    ABEF_SAVE_X = STATE0_X; \
-    ABEF_SAVE_Y = STATE0_Y; \
-    CDGH_SAVE_X = STATE1_X; \
-    CDGH_SAVE_Y = STATE1_Y; \
+    ABCD_SAVE_X = STATE0_X; \
+    ABCD_SAVE_Y = STATE0_Y; \
+    EFGH_SAVE_X = STATE1_X; \
+    EFGH_SAVE_Y = STATE1_Y; \
 \
     MSG0_X = load_msg( input_X, 0 ); \
     MSG0_Y = load_msg( input_Y, 0 ); \
@@ -1245,10 +1267,10 @@ void sha256_neon_sha_transform_le( uint32_t *state_out, const void *input,
     STATE0_Y = vsha256hq_u32( STATE0_Y, STATE1_Y, TMP1_Y ); \
     STATE1_X = vsha256h2q_u32( STATE1_X, TMP2_X, TMP1_X ); \
     STATE1_Y = vsha256h2q_u32( STATE1_Y, TMP2_Y, TMP1_Y ); \
-    STATE0_X = vaddq_u32( STATE0_X, ABEF_SAVE_X ); \
-    STATE0_Y = vaddq_u32( STATE0_Y, ABEF_SAVE_Y ); \
-    STATE1_X = vaddq_u32( STATE1_X, CDGH_SAVE_X ); \
-    STATE1_Y = vaddq_u32( STATE1_Y, CDGH_SAVE_Y ); \
+    STATE0_X = vaddq_u32( STATE0_X, ABCD_SAVE_X ); \
+    STATE0_Y = vaddq_u32( STATE0_Y, ABCD_SAVE_Y ); \
+    STATE1_X = vaddq_u32( STATE1_X, EFGH_SAVE_X ); \
+    STATE1_Y = vaddq_u32( STATE1_Y, EFGH_SAVE_Y ); \
     vst1q_u32( state_out_X  , STATE0_X ); \
     vst1q_u32( state_out_Y  , STATE0_Y ); \
     vst1q_u32( state_out_X+4, STATE1_X ); \

algo/sha/sha256-hash.h

@@ -5,27 +5,21 @@
 #include "simd-utils.h"
 #include "cpuminer-config.h"
 
-// generic interface 
+static const uint32_t SHA256_IV[8];
+
+#if defined(__x86_64__) && defined(__SHA__)
 
 typedef struct
 {
-   unsigned char buf[64];    /* first field, for alignment */
+   unsigned char buf[64];
    uint32_t state[8];
    uint64_t count;
 } sha256_context __attribute__((aligned(64)));
 
-static const uint32_t SHA256_IV[8];
-
 void sha256_full( void *hash, const void *data, size_t len );
 void sha256_update( sha256_context *ctx, const void *data, size_t len );
 void sha256_final( sha256_context *ctx, void *hash );
 void sha256_ctx_init( sha256_context *ctx );
-void sha256_transform_le( uint32_t *state_out, const uint32_t *data,
-                          const uint32_t *state_in );
-void sha256_transform_be( uint32_t *state_out, const uint32_t *data,
-                          const uint32_t *state_in );
-
-#if defined(__x86_64__) && defined(__SHA__)
 
 void sha256_x86_sha_transform_le( uint32_t *state_out, const void *input,
                                   const uint32_t *state_in );
@@ -50,14 +44,6 @@ void sha256_x86_x2sha_final_rounds( uint32_t *state_out_X, uint32_t *state_out_Y
                  const uint32_t *state_mid_X, const uint32_t *state_mid_Y,
                  const uint32_t *state_save_X, const uint32_t *state_save_Y );
 
-// Temporary during name transition
-#define sha256_opt_transform_le   sha256_x86_sha_transform_le
-#define sha256_opt_transform_be   sha256_x86_sha_transform_be
-#define sha256_ni2x_transform_le  sha256_x86_x2sha_transform_le
-#define sha256_ni2x_transform_be  sha256_x86_x2sha_transform_be
-#define sha256_ni_prehash_3rounds sha256_x86_sha_prehash_3rounds
-#define sha256_ni2x_final_rounds  sha256_x86_x2sha_final_rounds
-
 // generic API
 #define sha256_transform_le        sha256_x86_sha_transform_le
 #define sha256_transform_be        sha256_x86_sha_transform_be
@@ -68,6 +54,20 @@ void sha256_x86_x2sha_final_rounds( uint32_t *state_out_X, uint32_t *state_out_Y
 
 #elif defined(__ARM_NEON) && defined(__ARM_FEATURE_SHA2)
 
+// SHA-256 AArch64 with NEON & SHA2
+
+typedef struct
+{
+   unsigned char buf[64];
+   uint32_t state[8];
+   uint64_t count;
+} sha256_context __attribute__((aligned(64)));
+
+void sha256_full( void *hash, const void *data, size_t len );
+void sha256_update( sha256_context *ctx, const void *data, size_t len );
+void sha256_final( sha256_context *ctx, void *hash );
+void sha256_ctx_init( sha256_context *ctx );
+
 void sha256_neon_sha_transform_be( uint32_t *state_out, const void *input,
                                    const uint32_t *state_in );
 void sha256_neon_sha_transform_le( uint32_t *state_out, const void *input,
@@ -89,14 +89,6 @@ void sha256_neon_x2sha_final_rounds( uint32_t *state_out_X,
                  const uint32_t *state_mid_X, const uint32_t *state_mid_Y,
                  const uint32_t *state_save_X, const uint32_t *state_save_Y );
 
-// Temporary during name transition
-#define sha256_transform_le         sha256_neon_sha_transform_le
-#define sha256_transform_be         sha256_neon_sha_transform_be
-#define sha256_2x_transform_le      sha256_neon_x2sha_transform_le
-#define sha256_2x_transform_be      sha256_neon_x2sha_transform_be
-#define sha256_prehash_3rounds      sha256_neon_sha_prehash_3rounds
-#define sha256_2x_final_rounds      sha256_neon_x2sha_final_rounds
-
 // generic API
 #define sha256_transform_le        sha256_neon_sha_transform_le
 #define sha256_transform_be        sha256_neon_sha_transform_be
@@ -106,9 +98,11 @@ void sha256_neon_x2sha_final_rounds( uint32_t *state_out_X,
 #define sha256_2x_final_rounds     sha256_neon_x2sha_final_rounds
 
 #else
+
 // without HW acceleration...
 #include "sph_sha2.h"
 
+#define sha256_context              sph_sha256_context
 #define sha256_full                 sph_sha256_full
 #define sha256_ctx_init             sph_sha256_init
 #define sha256_update               sph_sha256
@@ -117,12 +111,11 @@ void sha256_neon_x2sha_final_rounds( uint32_t *state_out_X,
 #define sha256_transform_be         sph_sha256_transform_be
 #define sha256_prehash_3rounds      sph_sha256_prehash_3rounds
 
-
 #endif
 
-#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) && defined(__AVX512BW__)
+#if defined(SIMD512)
 
-// SHA-256 16 way
+// SHA-256 16 way x86_64
 
 typedef struct
 {
@@ -147,7 +140,7 @@ void sha256_16x32_final_rounds( __m512i *state_out, const __m512i *data,
 int sha256_16x32_transform_le_short( __m512i *state_out, const __m512i *data,
                             const __m512i *state_in, const uint32_t *target );
 
-#define sha256_16way_context sha256_16x32_context
+#define sha256_16way_context               sha256_16x32_context
 #define sha256_16way_init                  sha256_16x32_init
 #define sha256_16way_update                sha256_16x32_update
 #define sha256_16way_close                 sha256_16x32_close
@@ -162,7 +155,7 @@ int sha256_16x32_transform_le_short( __m512i *state_out, const __m512i *data,
 
 #if defined (__AVX2__)
 
-// SHA-256 8 way
+// SHA-256 8 way x86_64
 
 typedef struct
 {
@@ -201,7 +194,7 @@ int sha256_8x32_transform_le_short( __m256i *state_out, const __m256i *data,
 
 #endif  // AVX2
 
-// SHA-256 4 way
+// SHA-256 4 way x86_64 with SSE2 or AArch64 with NEON
 
 typedef struct
 {