Handle lyra2v3 algo, for VTC fork

mostly imported from opensourced vertcoin-miner with a few fixes

Makefile.am

@@ -38,6 +38,7 @@ ccminer_SOURCES	= elist.h miner.h compat.h \
 			  lyra2/Lyra2.c lyra2/Sponge.c \
 			  lyra2/lyra2RE.cu lyra2/cuda_lyra2.cu \
 		          lyra2/lyra2REv2.cu lyra2/cuda_lyra2v2.cu \
+			  lyra2/lyra2REv3.cu lyra2/cuda_lyra2v3.cu \
 			  lyra2/Lyra2Z.c lyra2/lyra2Z.cu lyra2/cuda_lyra2Z.cu \
 			  lyra2/allium.cu \
 			  Algo256/cuda_bmw256.cu Algo256/cuda_cubehash256.cu \

README.txt

@@ -1,5 +1,5 @@
 
-ccminer 2.3                     "phi2 and cryptonight variants"
+ccminer 2.3.1                     "lyra2v3, exosis and sha256q"
 ---------------------------------------------------------------
 
 ***************************************************************
@@ -100,7 +100,8 @@ its command line interface and options.
                           lbry        use to mine LBRY Credits
                           luffa       use to mine Joincoin
                           lyra2       use to mine CryptoCoin
-                          lyra2v2     use to mine Vertcoin
+                          lyra2v2     use to mine Monacoin
+                          lyra2v3     use to mine Vertcoin
                           lyra2z      use to mine Zerocoin (XZC)
                           monero      use to mine Monero (XMR)
                           myr-gr      use to mine Myriad-Groest
@@ -117,7 +118,7 @@ its command line interface and options.
                           scrypt-jane use to mine Chacha coins like Cache and Ultracoin
                           s3          use to mine 1coin (ONE)
                           sha256t     use to mine OneCoin (OC)
-			  sha256q     use to mine Pyrite
+                          sha256q     use to mine Pyrite
                           sia         use to mine SIA
                           sib         use to mine Sibcoin
                           skein       use to mine Skeincoin

algos.h

@@ -34,6 +34,7 @@ enum sha_algos {
 	ALGO_LUFFA,
 	ALGO_LYRA2,
 	ALGO_LYRA2v2,
+	ALGO_LYRA2v3,
 	ALGO_LYRA2Z,
 	ALGO_MJOLLNIR,		/* Hefty hash */
 	ALGO_MYR_GR,
@@ -115,6 +116,7 @@ static const char *algo_names[] = {
 	"luffa",
 	"lyra2",
 	"lyra2v2",
+	"lyra2v3",
 	"lyra2z",
 	"mjollnir",
 	"myr-gr",
@@ -199,6 +201,8 @@ static inline int algo_to_int(char* arg)
 			i = ALGO_LYRA2;
 		else if (!strcasecmp("lyra2rev2", arg))
 			i = ALGO_LYRA2v2;
+		else if (!strcasecmp("lyra2rev3", arg))
+			i = ALGO_LYRA2v3;
 		else if (!strcasecmp("phi1612", arg))
 			i = ALGO_PHI;
 		else if (!strcasecmp("bitcoin", arg))

bench.cpp

@@ -78,6 +78,7 @@ void algo_free_all(int thr_id)
 	free_luffa(thr_id);
 	free_lyra2(thr_id);
 	free_lyra2v2(thr_id);
+	free_lyra2v3(thr_id);
 	free_lyra2Z(thr_id);
 	free_myriad(thr_id);
 	free_neoscrypt(thr_id);

ccminer.cpp

@@ -269,7 +269,8 @@ Options:\n\
 			lbry        LBRY Credits (Sha/Ripemd)\n\
 			luffa       Joincoin\n\
 			lyra2       CryptoCoin\n\
-			lyra2v2     VertCoin\n\
+			lyra2v2     MonaCoin\n\
+			lyra2v3     Vertcoin\n\
 			lyra2z      ZeroCoin (3rd impl)\n\
 			myr-gr      Myriad-Groestl\n\
 			monero      XMR cryptonight (v7)\n\
@@ -1742,6 +1743,7 @@ static bool stratum_gen_work(struct stratum_ctx *sctx, struct work *work)
 		case ALGO_KECCAKC:
 		case ALGO_LBRY:
 		case ALGO_LYRA2v2:
+		case ALGO_LYRA2v3:
 		case ALGO_LYRA2Z:
 		case ALGO_PHI2:
 		case ALGO_TIMETRAVEL:
@@ -2283,6 +2285,7 @@ static void *miner_thread(void *userdata)
 			case ALGO_JHA:
 			case ALGO_HSR:
 			case ALGO_LYRA2v2:
+			case ALGO_LYRA2v3:
 			case ALGO_PHI:
 			case ALGO_PHI2:
 			case ALGO_POLYTIMOS:
@@ -2474,6 +2477,9 @@ static void *miner_thread(void *userdata)
 		case ALGO_LYRA2v2:
 			rc = scanhash_lyra2v2(thr_id, &work, max_nonce, &hashes_done);
 			break;
+		case ALGO_LYRA2v3:
+			rc = scanhash_lyra2v3(thr_id, &work, max_nonce, &hashes_done);
+			break;
 		case ALGO_LYRA2Z:
 			rc = scanhash_lyra2Z(thr_id, &work, max_nonce, &hashes_done);
 			break;

ccminer.vcxproj

@@ -530,6 +530,9 @@
     <ClInclude Include="lyra2\cuda_lyra2_sm2.cuh" />
     <ClInclude Include="lyra2\cuda_lyra2_sm5.cuh" />
     <ClInclude Include="lyra2\cuda_lyra2v2_sm3.cuh" />
+    <CudaCompile Include="lyra2\lyra2REv3.cu" />
+    <CudaCompile Include="lyra2\cuda_lyra2v3.cu" />
+    <ClInclude Include="lyra2\cuda_lyra2v3_sm3.cuh" />
     <CudaCompile Include="lyra2\lyra2Z.cu" />
     <CudaCompile Include="lyra2\cuda_lyra2Z.cu" />
     <ClInclude Include="lyra2\cuda_lyra2Z_sm5.cuh" />

ccminer.vcxproj.filters

@@ -946,6 +946,15 @@
     <CudaCompile Include="lyra2\lyra2REv2.cu">
       <Filter>Source Files\CUDA\lyra2</Filter>
     </CudaCompile>
+    <CudaCompile Include="lyra2\cuda_lyra2v3.cu">
+      <Filter>Source Files\CUDA\lyra2</Filter>
+    </CudaCompile>
+    <ClInclude Include="lyra2\cuda_lyra2v3_sm3.cuh">
+      <Filter>Source Files\CUDA\lyra2</Filter>
+    </CudaCompile>
+    <CudaCompile Include="lyra2\lyra2REv3.cu">
+      <Filter>Source Files\CUDA\lyra2</Filter>
+    </CudaCompile>
     <CudaCompile Include="lyra2\cuda_lyra2Z.cu">
       <Filter>Source Files\CUDA\lyra2</Filter>
     </CudaCompile>

compat/ccminer-config.h

@@ -164,7 +164,7 @@
 #define PACKAGE_URL "http://github.com/tpruvot/ccminer"
 
 /* Define to the version of this package. */
-#define PACKAGE_VERSION "2.3"
+#define PACKAGE_VERSION "2.3.1"
 
 /* If using the C implementation of alloca, define if you know the
    direction of stack growth for your system; otherwise it will be

lyra2/Lyra2.c

@@ -212,3 +212,176 @@ int LYRA2(void *K, int64_t kLen, const void *pwd, int32_t pwdlen, const void *sa
 
 	return 0;
 }
+
+int LYRA2_3(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)
+{
+	//============================= Basic variables ============================//
+	int64_t row = 2; //index of row to be processed
+	int64_t prev = 1; //index of prev (last row ever computed/modified)
+	int64_t rowa = 0; //index of row* (a previous row, deterministically picked during Setup and randomly picked while Wandering)
+	int64_t tau; //Time Loop iterator
+	int64_t step = 1; //Visitation step (used during Setup and Wandering phases)
+	int64_t window = 2; //Visitation window (used to define which rows can be revisited during Setup)
+	int64_t gap = 1; //Modifier to the step, assuming the values 1 or -1
+	int64_t i; //auxiliary iteration counter
+	int64_t v64; // 64bit var for memcpy
+	uint64_t instance = 0;
+	//==========================================================================/
+
+	//========== Initializing the Memory Matrix and pointers to it =============//
+	//Tries to allocate enough space for the whole memory matrix
+
+	const int64_t ROW_LEN_INT64 = BLOCK_LEN_INT64 * nCols;
+	const int64_t ROW_LEN_BYTES = ROW_LEN_INT64 * 8;
+	// for Lyra2REv2, nCols = 4, v1 was using 8
+	const int64_t BLOCK_LEN = (nCols == 4) ? BLOCK_LEN_BLAKE2_SAFE_INT64 : BLOCK_LEN_BLAKE2_SAFE_BYTES;
+
+	size_t sz = (size_t)ROW_LEN_BYTES * nRows;
+	uint64_t *wholeMatrix = malloc(sz);
+	if (wholeMatrix == NULL) {
+		return -1;
+	}
+	memset(wholeMatrix, 0, sz);
+
+	//Allocates pointers to each row of the matrix
+	uint64_t **memMatrix = malloc(sizeof(uint64_t*) * nRows);
+	if (memMatrix == NULL) {
+		return -1;
+	}
+	//Places the pointers in the correct positions
+	uint64_t *ptrWord = wholeMatrix;
+	for (i = 0; i < nRows; i++) {
+		memMatrix[i] = ptrWord;
+		ptrWord += ROW_LEN_INT64;
+	}
+	//==========================================================================/
+
+	//============= Getting the password + salt + basil padded with 10*1 ===============//
+	//OBS.:The memory matrix will temporarily hold the password: not for saving memory,
+	//but this ensures that the password copied locally will be overwritten as soon as possible
+
+	//First, we clean enough blocks for the password, salt, basil and padding
+	int64_t nBlocksInput = ((saltlen + pwdlen + 6 * sizeof(uint64_t)) / BLOCK_LEN_BLAKE2_SAFE_BYTES) + 1;
+
+	byte *ptrByte = (byte*) wholeMatrix;
+
+	//Prepends the password
+	memcpy(ptrByte, pwd, pwdlen);
+	ptrByte += pwdlen;
+
+	//Concatenates the salt
+	memcpy(ptrByte, salt, saltlen);
+	ptrByte += saltlen;
+
+	memset(ptrByte, 0, (size_t) (nBlocksInput * BLOCK_LEN_BLAKE2_SAFE_BYTES - (saltlen + pwdlen)));
+
+	//Concatenates the basil: every integer passed as parameter, in the order they are provided by the interface
+	memcpy(ptrByte, &kLen, sizeof(int64_t));
+	ptrByte += sizeof(uint64_t);
+	v64 = pwdlen;
+	memcpy(ptrByte, &v64, sizeof(int64_t));
+	ptrByte += sizeof(uint64_t);
+	v64 = saltlen;
+	memcpy(ptrByte, &v64, sizeof(int64_t));
+	ptrByte += sizeof(uint64_t);
+	v64 = timeCost;
+	memcpy(ptrByte, &v64, sizeof(int64_t));
+	ptrByte += sizeof(uint64_t);
+	v64 = nRows;
+	memcpy(ptrByte, &v64, sizeof(int64_t));
+	ptrByte += sizeof(uint64_t);
+	v64 = nCols;
+	memcpy(ptrByte, &v64, sizeof(int64_t));
+	ptrByte += sizeof(uint64_t);
+
+	//Now comes the padding
+	*ptrByte = 0x80; //first byte of padding: right after the password
+	ptrByte = (byte*) wholeMatrix; //resets the pointer to the start of the memory matrix
+	ptrByte += nBlocksInput * BLOCK_LEN_BLAKE2_SAFE_BYTES - 1; //sets the pointer to the correct position: end of incomplete block
+	*ptrByte ^= 0x01; //last byte of padding: at the end of the last incomplete block
+	//==========================================================================/
+
+	//======================= Initializing the Sponge State ====================//
+	//Sponge state: 16 uint64_t, BLOCK_LEN_INT64 words of them for the bitrate (b) and the remainder for the capacity (c)
+	uint64_t state[16];
+	initState(state);
+	//==========================================================================/
+
+	//================================ Setup Phase =============================//
+	//Absorbing salt, password and basil: this is the only place in which the block length is hard-coded to 512 bits
+	ptrWord = wholeMatrix;
+	for (i = 0; i < nBlocksInput; i++) {
+		absorbBlockBlake2Safe(state, ptrWord); //absorbs each block of pad(pwd || salt || basil)
+		ptrWord += BLOCK_LEN; //goes to next block of pad(pwd || salt || basil)
+	}
+
+	//Initializes M[0] and M[1]
+	reducedSqueezeRow0(state, memMatrix[0], nCols); //The locally copied password is most likely overwritten here
+
+	reducedDuplexRow1(state, memMatrix[0], memMatrix[1], nCols);
+
+	do {
+		//M[row] = rand; //M[row*] = M[row*] XOR rotW(rand)
+
+		reducedDuplexRowSetup(state, memMatrix[prev], memMatrix[rowa], memMatrix[row], nCols);
+
+		//updates the value of row* (deterministically picked during Setup))
+		rowa = (rowa + step) & (window - 1);
+		//update prev: it now points to the last row ever computed
+		prev = row;
+		//updates row: goes to the next row to be computed
+		row++;
+
+		//Checks if all rows in the window where visited.
+		if (rowa == 0) {
+		step = window + gap; //changes the step: approximately doubles its value
+		window *= 2; //doubles the size of the re-visitation window
+		gap = -gap; //inverts the modifier to the step
+	}
+
+	} while (row < nRows);
+	//==========================================================================/
+
+	//============================ Wandering Phase =============================//
+	row = 0; //Resets the visitation to the first row of the memory matrix
+	for (tau = 1; tau <= timeCost; tau++) {
+		//Step is approximately half the number of all rows of the memory matrix for an odd tau; otherwise, it is -1
+		step = ((tau & 1) == 0) ? -1 : (nRows >> 1) - 1;
+		do {
+			//Selects a pseudorandom index row* (the only change in REv3)
+			//------------------------------------------------------------------------------------------
+			instance = state[instance & 0xF];
+			rowa = state[instance & 0xF] & (unsigned int)(nRows-1);
+
+			//rowa = state[0] & (unsigned int)(nRows-1);  //(USE THIS IF nRows IS A POWER OF 2)
+			//rowa = state[0] % nRows; //(USE THIS FOR THE "GENERIC" CASE)
+			//------------------------------------------------------------------------------------------
+
+			//Performs a reduced-round duplexing operation over M[row*] XOR M[prev], updating both M[row*] and M[row]
+			reducedDuplexRow(state, memMatrix[prev], memMatrix[rowa], memMatrix[row], nCols);
+
+			//update prev: it now points to the last row ever computed
+			prev = row;
+
+			//updates row: goes to the next row to be computed
+			//------------------------------------------------------------------------------------------
+			row = (row + step) & (unsigned int)(nRows-1); //(USE THIS IF nRows IS A POWER OF 2)
+			//row = (row + step) % nRows; //(USE THIS FOR THE "GENERIC" CASE)
+			//------------------------------------------------------------------------------------------
+
+		} while (row != 0);
+	}
+
+	//============================ Wrap-up Phase ===============================//
+	//Absorbs the last block of the memory matrix
+	absorbBlock(state, memMatrix[rowa]);
+
+	//Squeezes the key
+	squeeze(state, K, (unsigned int) kLen);
+
+	//========================= Freeing the memory =============================//
+	free(memMatrix);
+	free(wholeMatrix);
+
+	return 0;
+}

lyra2/Lyra2.h

@@ -38,5 +38,6 @@ typedef unsigned char byte;
 #endif
 
 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);
+int LYRA2_3(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);
 
 #endif /* LYRA2_H_ */