v23.11

RELEASE_NOTES

@@ -27,17 +27,19 @@ See INSTALL_LINUX or INSTALL_WINDOWS for compile instructions
 Requirements
 ------------
 
-Intel Core2 or newer, or AMD Steamroller or newer CPU. ARM CPUs are not
-supported.
+- A x86_64 architecture CPU with a minimum of SSE2 support. This includes Intel Core2 and newer and AMD equivalents.
+- Arm CPU supporting AArch64 and NEON.
 
-64 bit Linux or Windows operating system. Apple, Android and Raspberry Pi
-are not supported. FreeBSD YMMV.
+32 bit CPUs are not supported.
 
-ARM requirements (Beta):
+Older CPUs are supported by open source cpuminer-multi by TPruvot but at reduced performance.
 
-CPU: Armv8 and NEON, SHA2 & AES are optional
-OS: Linux distribution built for AArch64.
-Packages: source code only.
+Mining on mobile devices that meet the requirements is not recommended due to the risk of
+overheating and damaging the battery. Mining has unlimited demand, it will push any device
+to or beyond its limits. There is also a fire risk with overheated lithium batteries.
+
+Beware of apps claiming "mobile only mining". There is no such thing, they aren't miners.
+If a mobile CPU can mine it any CPU can.
 
 See wiki for details.
 
@@ -73,12 +75,18 @@ If not what makes it happen or not happen?
 Change Log
 ----------
 
+v23.11
+
+This is a release candidate for full AArch64 support, marking the end of the Beta phase.
+Fixed hmq1725 & x25x algos, SSE2 & NEON, broken in v3.23.4.
+Most CPU-mineable SHA3 algos (X*) upgraded to 2-way SSE2 & NEON.
+
 v23.10
 
 x86_64: Fixed scrypt, scryptn2 algos SSE2. 
-Fixed sha512d256d algo AVX2, SSE2, NEON.
+Fixed sha512256d algo AVX2, SSE2, NEON.
 Fixed a bug in Skein N-way that reduced performance.
-ARM: Skein algo optimized for NEON & SHA2.
+ARM: Skein optimized for NEON, SHA2 & SSE2.
 Skein2 algo 2-way optimized for NEON & SSE2.
 
 v23.9

algo-gate-api.h

@@ -99,7 +99,7 @@ typedef  uint32_t set_t;
 #define AES_OPT          1 <<  7   // Intel Westmere, AArch64
 #define VAES_OPT         1 <<  8   // Icelake, Zen3
 #define SHA_OPT          1 <<  9   // Zen1, Icelake, AArch64 
-#define SHA512_OPT       1 << 10   // AArch64 
+#define SHA512_OPT       1 << 10   // Intel Arrow Lake, AArch64 
 #define NEON_OPT         1 << 11   // AArch64 
 
 // AVX10 does not have explicit algo features:

algo/bmw/bmw512-4way.c

@@ -2,12 +2,11 @@
 #include <stdlib.h>
 #include <string.h>
 #include <stdint.h>
-//#include "sph_keccak.h"
 #include "bmw-hash-4way.h"
 
 #if defined(BMW512_8WAY)
 
-void bmw512hash_8way(void *state, const void *input)
+void bmw512hash_8way( void *state, const void *input )
 {
     bmw512_8way_context ctx;
     bmw512_8way_init( &ctx );
@@ -27,9 +26,9 @@ int scanhash_bmw512_8way( struct work *work, uint32_t max_nonce,
    uint32_t n = pdata[19];
    const uint32_t first_nonce = pdata[19];
    const uint32_t last_nonce = max_nonce - 8;
-   __m512i  *noncev = (__m512i*)vdata + 9;   // aligned
+   __m512i  *noncev = (__m512i*)vdata + 9;
    const uint32_t Htarg = ptarget[7];
-   int thr_id = mythr->id;
+   const int thr_id = mythr->id;
 
    mm512_bswap32_intrlv80_8x64( vdata, pdata );
    do {
@@ -43,7 +42,7 @@ int scanhash_bmw512_8way( struct work *work, uint32_t max_nonce,
       if ( unlikely( hash7[ lane<<1 ] <= Htarg ) )
       {
           extr_lane_8x64( lane_hash, hash, lane, 256 );
-          if ( fulltest( lane_hash, ptarget ) )
+          if ( likely( valid_hash( lane_hash, ptarget ) && !opt_benchmark ))
           {
               pdata[19] = n + lane;
               submit_solution( work, lane_hash, mythr );
@@ -59,9 +58,7 @@ int scanhash_bmw512_8way( struct work *work, uint32_t max_nonce,
 
 #elif defined(BMW512_4WAY)
 
-//#ifdef BMW512_4WAY
-
-void bmw512hash_4way(void *state, const void *input)
+void bmw512hash_4way( void *state, const void *input )
 {
     bmw512_4way_context ctx;
     bmw512_4way_init( &ctx );
@@ -80,10 +77,10 @@ int scanhash_bmw512_4way( struct work *work, uint32_t max_nonce,
    uint32_t *ptarget = work->target;
    uint32_t n = pdata[19];
    const uint32_t first_nonce = pdata[19];
-   const uint32_t last_nonce = max_nonce -  4;
-   __m256i  *noncev = (__m256i*)vdata + 9;   // aligned
+   const uint32_t last_nonce = max_nonce - 4;
+   __m256i  *noncev = (__m256i*)vdata + 9; 
    const uint32_t Htarg = ptarget[7];
-    int thr_id = mythr->id;  // thr_id arg is deprecated
+   const int thr_id = mythr->id;  
 
    mm256_bswap32_intrlv80_4x64( vdata, pdata );
    do {
@@ -96,7 +93,7 @@ int scanhash_bmw512_4way( struct work *work, uint32_t max_nonce,
       if ( unlikely( hash7[ lane<<1 ] <= Htarg ) )
       {
           extr_lane_4x64( lane_hash, hash, lane, 256 );
-          if ( fulltest( lane_hash, ptarget ) )
+          if ( likely( valid_hash( lane_hash, ptarget ) && !opt_benchmark ))
           {
               pdata[19] = n + lane;
               submit_solution( work, lane_hash, mythr );
@@ -110,4 +107,55 @@ int scanhash_bmw512_4way( struct work *work, uint32_t max_nonce,
    return 0;
 }
 
+#elif defined(BMW512_2WAY)
+
+void bmw512hash_2x64( void *state, const void *input )
+{
+    bmw512_2x64_context ctx;
+    bmw512_2x64_init( &ctx );
+    bmw512_2x64_update( &ctx, input, 80 );
+    bmw512_2x64_close( &ctx, state );
+}
+
+int scanhash_bmw512_2x64( struct work *work, uint32_t max_nonce,
+                          uint64_t *hashes_done, struct thr_info *mythr )
+{
+   uint32_t vdata[24*2] __attribute__ ((aligned (64)));
+   uint32_t hash[16*2] __attribute__ ((aligned (32)));
+   uint32_t lane_hash[8] __attribute__ ((aligned (32)));
+   uint32_t *hash7 = &(hash[13]);   // 3*4+1
+   uint32_t *pdata = work->data;
+   uint32_t *ptarget = work->target;
+   uint32_t n = pdata[19];
+   const uint32_t first_nonce = pdata[19];
+   const uint32_t last_nonce = max_nonce - 2;
+   v128_t *noncev = (v128_t*)vdata + 9;  
+   const uint32_t Htarg = ptarget[7];
+   const int thr_id = mythr->id; 
+
+   v128_bswap32_intrlv80_2x64( vdata, pdata );
+   do {
+      *noncev = v128_intrlv_blend_32( v128_bswap32(
+                                      v128_set32( n+1, 0, n, 0 ) ), *noncev );
+
+      bmw512hash_2x64( hash, vdata );
+
+      for ( int lane = 0; lane < 2; lane++ )
+      if ( unlikely( hash7[ lane<<1 ] <= Htarg ) )
+      {
+          extr_lane_2x64( lane_hash, hash, lane, 256 );
+          if ( likely( valid_hash( lane_hash, ptarget ) && !opt_benchmark ))
+          {
+              pdata[19] = n + lane;
+              submit_solution( work, lane_hash, mythr );
+          }
+      }
+      n += 2;
+
+   } while ( likely( (n < last_nonce) && !work_restart[thr_id].restart ) );
+
+   *hashes_done = n - first_nonce;
+   return 0;
+}
+
 #endif

algo/bmw/bmw512-gate.c

@@ -2,14 +2,17 @@
 
 bool register_bmw512_algo( algo_gate_t* gate )
 {
-  gate->optimizations = AVX2_OPT | AVX512_OPT;
+  gate->optimizations = SSE2_OPT | AVX2_OPT | AVX512_OPT | NEON_OPT;
   opt_target_factor = 256.0;
 #if defined (BMW512_8WAY)
   gate->scanhash  = (void*)&scanhash_bmw512_8way;
   gate->hash      = (void*)&bmw512hash_8way;
 #elif defined (BMW512_4WAY)
   gate->scanhash  = (void*)&scanhash_bmw512_4way;
   gate->hash      = (void*)&bmw512hash_4way;
+#elif defined (BMW512_2WAY)
+  gate->scanhash  = (void*)&scanhash_bmw512_2x64;
+  gate->hash      = (void*)&bmw512hash_2x64;
 #else
   gate->scanhash        = (void*)&scanhash_bmw512;
   gate->hash            = (void*)&bmw512hash;

algo/bmw/bmw512-gate.h

@@ -8,19 +8,27 @@
   #define BMW512_8WAY 1
 #elif defined(__AVX2__)
   #define BMW512_4WAY 1
+#elif defined(__SSE2__) || defined(__ARM_NEON)
+  #define BMW512_2WAY 1
 #endif
 
 #if defined(BMW512_8WAY)
 
 void bmw512hash_8way( void *state, const void *input );
 int scanhash_bmw512_8way( struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done, struct thr_info *mythr );
+                          uint64_t *hashes_done, struct thr_info *mythr );
 
 #elif defined(BMW512_4WAY)
 
 void bmw512hash_4way( void *state, const void *input );
 int scanhash_bmw512_4way( struct work *work, uint32_t max_nonce,
-                         uint64_t *hashes_done, struct thr_info *mythr );
+                          uint64_t *hashes_done, struct thr_info *mythr );
+
+#elif defined(BMW512_2WAY)
+
+void bmw512hash_2x64( void *state, const void *input );
+int scanhash_bmw512_2x64( struct work *work, uint32_t max_nonce,
+                          uint64_t *hashes_done, struct thr_info *mythr );
 
 #else
 

algo/echo/aes_ni/hash.c

@@ -236,9 +236,7 @@ void Compress(hashState_echo *ctx, const unsigned char *pmsg, unsigned int uBloc
 
 }
 
-
-
-HashReturn init_echo(hashState_echo *ctx, int nHashSize)
+HashReturn init_echo( hashState_echo *ctx, int nHashSize )
 {
 	int i, j;
 
@@ -280,7 +278,8 @@ HashReturn init_echo(hashState_echo *ctx, int nHashSize)
 	return SUCCESS;
 }
 
-HashReturn update_echo(hashState_echo *state, const BitSequence *data, DataLength databitlen)
+HashReturn update_echo( hashState_echo *state, const void *data,
+                        uint32_t databitlen )
 {
 	unsigned int uByteLength, uBlockCount, uRemainingBytes;
 
@@ -330,7 +329,7 @@ HashReturn update_echo(hashState_echo *state, const BitSequence *data, DataLengt
 	return SUCCESS;
 }
 
-HashReturn final_echo(hashState_echo *state, BitSequence *hashval)
+HashReturn final_echo( hashState_echo *state, void *hashval)
 {
 	v128_t remainingbits;
 
@@ -407,8 +406,8 @@ HashReturn final_echo(hashState_echo *state, BitSequence *hashval)
 	return SUCCESS;
 }
 
-HashReturn update_final_echo( hashState_echo *state, BitSequence *hashval,
-                              const BitSequence *data, DataLength databitlen )
+HashReturn update_final_echo( hashState_echo *state, void *hashval,
+                              const void *data, uint32_t databitlen )
 {
    unsigned int uByteLength, uBlockCount, uRemainingBytes;
 
@@ -530,8 +529,8 @@ HashReturn update_final_echo( hashState_echo *state, BitSequence *hashval,
    return SUCCESS;
 }
 
-HashReturn echo_full( hashState_echo *state, BitSequence *hashval,
-            int nHashSize, const BitSequence *data, DataLength datalen )
+HashReturn echo_full( hashState_echo *state, void *hashval,
+            int nHashSize, const void *data, uint32_t datalen )
 {
    int i, j;
 
@@ -578,7 +577,7 @@ HashReturn echo_full( hashState_echo *state, BitSequence *hashval,
         {
            // Fill the buffer
            memcpy( state->buffer + state->uBufferBytes,
-                   (void*)data, state->uBlockLength - state->uBufferBytes );
+                   data, state->uBlockLength - state->uBufferBytes );
 
            // Process buffer
            Compress( state, state->buffer, 1 );
@@ -601,7 +600,7 @@ HashReturn echo_full( hashState_echo *state, BitSequence *hashval,
         }
 
         if( uRemainingBytes > 0 )
-        memcpy(state->buffer, (void*)data, uRemainingBytes);
+        memcpy(state->buffer, data, uRemainingBytes);
 
         state->uBufferBytes = uRemainingBytes;
    }
@@ -689,7 +688,7 @@ HashReturn echo_full( hashState_echo *state, BitSequence *hashval,
 }
 
 
-
+#if 0
 HashReturn hash_echo(int hashbitlen, const BitSequence *data, DataLength databitlen, BitSequence *hashval)
 {
 	HashReturn hRet;
@@ -746,5 +745,6 @@ HashReturn hash_echo(int hashbitlen, const BitSequence *data, DataLength databit
 
 	return SUCCESS;
 }
+#endif
 
 #endif

algo/echo/aes_ni/hash_api.h

@@ -47,16 +47,16 @@ HashReturn init_echo(hashState_echo *state, int hashbitlen);
 
 HashReturn reinit_echo(hashState_echo *state);
 
-HashReturn update_echo(hashState_echo *state, const BitSequence *data, DataLength databitlen);
+HashReturn update_echo(hashState_echo *state, const void *data, uint32_t databitlen);
 
-HashReturn final_echo(hashState_echo *state, BitSequence *hashval);
+HashReturn final_echo(hashState_echo *state, void *hashval);
 
-HashReturn hash_echo(int hashbitlen, const BitSequence *data, DataLength databitlen, BitSequence *hashval);
+HashReturn hash_echo(int hashbitlen, const void *data, uint32_t databitlen, void *hashval);
 
-HashReturn update_final_echo( hashState_echo *state, BitSequence *hashval,
-                              const BitSequence *data, DataLength databitlen );
-HashReturn echo_full( hashState_echo *state, BitSequence *hashval,
-            int nHashSize, const BitSequence *data, DataLength databitlen );
+HashReturn update_final_echo( hashState_echo *state, void *hashval,
+                              const void *data, uint32_t databitlen );
+HashReturn echo_full( hashState_echo *state, void *hashval,
+            int nHashSize, const void *data, uint32_t databitlen );
 
 #endif // HASH_API_H
 

algo/echo/sph_echo.c

@@ -36,7 +36,6 @@
 
 #include "sph_echo.h"
 
-#if !defined(__AES__)
 
 #ifdef __cplusplus
 extern "C"{
@@ -1031,4 +1030,3 @@ sph_echo512_addbits_and_close(void *cc, unsigned ub, unsigned n, void *dst)
 #ifdef __cplusplus
 }
 #endif 
-#endif  // !AES

algo/echo/sph_echo.h

@@ -36,8 +36,6 @@
 #ifndef SPH_ECHO_H__
 #define SPH_ECHO_H__
 
-#if !defined(__AES__)
-
 #ifdef __cplusplus
 extern "C"{
 #endif
@@ -318,5 +316,4 @@ void sph_echo512_addbits_and_close(
 #ifdef __cplusplus
 }
 #endif
-#endif // !AES
 #endif

algo/groestl/sph_groestl.c

@@ -35,8 +35,6 @@
 
 #include "sph_groestl.h"
 
-#if !defined(__AES__)
-
 #ifdef __cplusplus
 extern "C"{
 #endif
@@ -3119,5 +3117,4 @@ sph_groestl512_addbits_and_close(void *cc, unsigned ub, unsigned n, void *dst)
 #ifdef __cplusplus
 }
 
-#endif  // !AES
 #endif

algo/groestl/sph_groestl.h

@@ -42,7 +42,6 @@ extern "C"{
 #include <stddef.h>
 #include "compat/sph_types.h"
 
-#if !defined(__AES__)   
 /**
  * Output size (in bits) for Groestl-224.
  */
@@ -327,5 +326,4 @@ void sph_groestl512_addbits_and_close(
 }
 #endif
 
-#endif  // !AES
 #endif