blake32.cu

/**
 * Blake-256 Cuda Kernel (Tested on SM 5.0)
 *
 * Tanguy Pruvot - Nov. 2014
 */

#define PRECALC64 1

#include "miner.h"

extern "C" {
#include "sph/sph_blake.h"
#include <stdint.h>
#include <memory.h>
}

/* threads per block and throughput (intensity) */
#define TPB 128

/* added in sph_blake.c */
extern "C" int blake256_rounds = 14;

/* hash by cpu with blake 256 */
extern "C" void blake256hash(void *output, const void *input, int8_t rounds = 14)
{
	uchar hash[64];
	sph_blake256_context ctx;

	blake256_rounds = rounds;

	sph_blake256_init(&ctx);
	sph_blake256(&ctx, input, 80);
	sph_blake256_close(&ctx, hash);

	memcpy(output, hash, 32);
}

#include "cuda_helper.h"

#if PRECALC64
__constant__ uint32_t _ALIGN(32) d_data[12];
#else
__constant__ static uint32_t _ALIGN(32) c_data[20];
/* midstate hash cache, this algo is run on 2 parts */
__device__ static uint32_t cache[8];
__device__ static uint32_t prevsum = 0;
/* crc32.c */
extern "C" uint32_t crc32_u32t(const uint32_t *buf, size_t size);
#endif

/* 8 adapters max (-t threads) */
static uint32_t *d_resNonce[8];
static uint32_t *h_resNonce[8];

/* max count of found nonces in one call */
#define NBN 2
static uint32_t extra_results[NBN] = { UINT32_MAX };

/* prefer uint32_t to prevent size conversions = speed +5/10 % */
__constant__
static uint32_t _ALIGN(32) c_sigma[16][16] = {
	{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
	{14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 },
	{11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 },
	{ 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 },
	{ 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 },
	{ 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 },
	{12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11 },
	{13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10 },
	{ 6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5 },
	{10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13 , 0 },
	{ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 },
	{14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3 },
	{11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4 },
	{ 7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8 },
	{ 9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13 },
	{ 2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9 }
};

#if !PRECALC64
__device__ __constant__
static const uint32_t __align__(32) c_IV256[8] = {
	SPH_C32(0x6A09E667), SPH_C32(0xBB67AE85),
	SPH_C32(0x3C6EF372), SPH_C32(0xA54FF53A),
	SPH_C32(0x510E527F), SPH_C32(0x9B05688C),
	SPH_C32(0x1F83D9AB), SPH_C32(0x5BE0CD19)
};
#endif

__device__ __constant__
static const uint32_t __align__(32) c_u256[16] = {
	SPH_C32(0x243F6A88), SPH_C32(0x85A308D3),
	SPH_C32(0x13198A2E), SPH_C32(0x03707344),
	SPH_C32(0xA4093822), SPH_C32(0x299F31D0),
	SPH_C32(0x082EFA98), SPH_C32(0xEC4E6C89),
	SPH_C32(0x452821E6), SPH_C32(0x38D01377),
	SPH_C32(0xBE5466CF), SPH_C32(0x34E90C6C),
	SPH_C32(0xC0AC29B7), SPH_C32(0xC97C50DD),
	SPH_C32(0x3F84D5B5), SPH_C32(0xB5470917)
};

#define GS(a,b,c,d,x) { \
	const uint32_t idx1 = c_sigma[r][x]; \
	const uint32_t idx2 = c_sigma[r][x+1]; \
	v[a] += (m[idx1] ^ c_u256[idx2]) + v[b]; \
	v[d] = SPH_ROTL32(v[d] ^ v[a], 16); \
	v[c] += v[d]; \
	v[b] = SPH_ROTR32(v[b] ^ v[c], 12); \
\
	v[a] += (m[idx2] ^ c_u256[idx1]) + v[b]; \
	v[d] = SPH_ROTR32(v[d] ^ v[a], 8); \
	v[c] += v[d]; \
	v[b] = SPH_ROTR32(v[b] ^ v[c], 7); \
}

/* Second part (64-80) msg never change, store it */
__device__ __constant__
static const uint32_t __align__(32) c_Padding[16] = {
	0, 0, 0, 0,
	0x80000000UL, 0, 0, 0,
	0, 0, 0, 0,
	0, 1, 0, 640,
};

__device__ static
void blake256_compress(uint32_t *h, const uint32_t *block, const uint32_t T0, const int rounds)
{
	uint32_t /*_ALIGN(8)*/ m[16];
	uint32_t v[16];

	m[0] = block[0];
	m[1] = block[1];
	m[2] = block[2];
	m[3] = block[3];

	for (int i = 4; i < 16; i++) {
#if PRECALC64
		m[i] = c_Padding[i];
#else
		m[i] = (T0 == 0x200) ? block[i] : c_Padding[i];
#endif
	}

	//#pragma unroll 8
	for(uint32_t i = 0; i < 8; i++)
		v[i] = h[i];

	v[ 8] = c_u256[0];
	v[ 9] = c_u256[1];
	v[10] = c_u256[2];
	v[11] = c_u256[3];

	v[12] = c_u256[4] ^ T0;
	v[13] = c_u256[5] ^ T0;
	v[14] = c_u256[6];
	v[15] = c_u256[7];

	for (int r = 0; r < rounds; r++) {
		/* column step */
		GS(0, 4, 0x8, 0xC, 0x0);
		GS(1, 5, 0x9, 0xD, 0x2);
		GS(2, 6, 0xA, 0xE, 0x4);
		GS(3, 7, 0xB, 0xF, 0x6);
		/* diagonal step */
		GS(0, 5, 0xA, 0xF, 0x8);
		GS(1, 6, 0xB, 0xC, 0xA);
		GS(2, 7, 0x8, 0xD, 0xC);
		GS(3, 4, 0x9, 0xE, 0xE);
	}
#if PRECALC64
	// only compute h6 & 7
	h[6U] ^= v[6U] ^ v[14U];
	h[7U] ^= v[7U] ^ v[15U];
#else
	//#pragma unroll 16
	for (int i = 0; i < 16; i++) {
		uint32_t j = i % 8U;
		h[j] ^= v[i];
	}
#endif
}

#if !PRECALC64 /* original method */
__global__
void blake256_gpu_hash_80(const uint32_t threads, const uint32_t startNonce, uint32_t *resNonce,
	const uint64_t highTarget, const int crcsum, const int rounds)
{
	uint32_t thread = (blockDim.x * blockIdx.x + threadIdx.x);
	if (thread < threads)
	{
		const uint32_t nonce = startNonce + thread;
		uint32_t h[8];

		#pragma unroll
		for(int i=0; i<8; i++) {
			h[i] = c_IV256[i];
		}

		if (crcsum != prevsum) {
			prevsum = crcsum;
			blake256_compress(h, c_data, 512, rounds);
			#pragma unroll
			for(int i=0; i<8; i++) {
				cache[i] = h[i];
			}
		} else {
			#pragma unroll
			for(int i=0; i<8; i++) {
				h[i] = cache[i];
			}
		}

		// ------ Close: Bytes 64 to 80 ------

		uint32_t ending[4];
		ending[0] = c_data[16];
		ending[1] = c_data[17];
		ending[2] = c_data[18];
		ending[3] = nonce; /* our tested value */

		blake256_compress(h, ending, 640, rounds);

		// not sure why, h[7] is ok
		h[6] = cuda_swab32(h[6]);

		// compare count of leading zeros h[6] + h[7]
		uint64_t high64 = ((uint64_t*)h)[3];
		if (high64 <= highTarget)
#if NBN == 2
		/* keep the smallest nonce, + extra one if found */
		if (resNonce[0] > nonce) {
			// printf("%llx %llx \n", high64, highTarget);
			resNonce[1] = resNonce[0];
			resNonce[0] = nonce;
		}
		else
			resNonce[1] = nonce;
#else
		resNonce[0] = nonce;
#endif
	}
}

__host__
uint32_t blake256_cpu_hash_80(const int thr_id, const uint32_t threads, const uint32_t startNonce, const uint64_t highTarget,
	const uint32_t crcsum, const int8_t rounds)
{
	const int threadsperblock = TPB;
	uint32_t result = UINT32_MAX;

	dim3 grid((threads + threadsperblock-1)/threadsperblock);
	dim3 block(threadsperblock);
	size_t shared_size = 0;

	/* Check error on Ctrl+C or kill to prevent segfaults on exit */
	if (cudaMemset(d_resNonce[thr_id], 0xff, NBN*sizeof(uint32_t)) != cudaSuccess)
		return result;

	blake256_gpu_hash_80<<<grid, block, shared_size>>>(threads, startNonce, d_resNonce[thr_id], highTarget, crcsum, (int) rounds);
	cudaDeviceSynchronize();
	if (cudaSuccess == cudaMemcpy(h_resNonce[thr_id], d_resNonce[thr_id], NBN*sizeof(uint32_t), cudaMemcpyDeviceToHost)) {
		//cudaThreadSynchronize(); /* seems no more required */
		result = h_resNonce[thr_id][0];
		for (int n=0; n < (NBN-1); n++)
			extra_results[n] = h_resNonce[thr_id][n+1];
	}
	return result;
}

__host__
void blake256_cpu_setBlock_80(uint32_t *pdata, const uint32_t *ptarget)
{
	uint32_t data[20];
	memcpy(data, pdata, 80);
	CUDA_SAFE_CALL(cudaMemcpyToSymbol(c_data, data, sizeof(data), 0, cudaMemcpyHostToDevice));
}
#else

/* ############################################################################################################################### */
/* Precalculated 1st 64-bytes block (midstate) method */

__global__
void blake256_gpu_hash_16(const uint32_t threads, const uint32_t startNonce, uint32_t *resNonce,
	const uint64_t highTarget, const int rounds, const bool trace)
{
	uint32_t thread = (blockDim.x * blockIdx.x + threadIdx.x);
	if (thread < threads)
	{
		const uint32_t nonce = startNonce + thread;
		uint32_t _ALIGN(16) h[8];

		#pragma unroll
		for(int i=0; i < 8; i++) {
			h[i] = d_data[i];
		}

		// ------ Close: Bytes 64 to 80 ------

		uint32_t _ALIGN(16) ending[4];
		ending[0] = d_data[8];
		ending[1] = d_data[9];
		ending[2] = d_data[10];
		ending[3] = nonce; /* our tested value */

		blake256_compress(h, ending, 640, rounds);

		//if (h[7] == 0 && high64 <= highTarget) {
		if (h[7] == 0) {
#if NBN == 2
			/* keep the smallest nonce, + extra one if found */
			if (resNonce[0] > nonce) {
				// printf("%llx %llx \n", high64, highTarget);
				resNonce[1] = resNonce[0];
				resNonce[0] = nonce;
			}
			else
				resNonce[1] = nonce;
#else
			resNonce[0] = nonce;
#endif
			if (trace) {
#ifdef _DEBUG
				uint64_t high64 = ((uint64_t*)h)[3];
				printf("gpu:  %16llx\n", high64);
				printf("gpu: %08x.%08x\n", h[7], h[6]);
				printf("tgt:  %16llx\n", highTarget);
#endif
			}
		}
	}
}

__host__
static uint32_t blake256_cpu_hash_16(const int thr_id, const uint32_t threads, const uint32_t startNonce, const uint64_t highTarget,
	const int8_t rounds)
{
	const int threadsperblock = TPB;
	uint32_t result = UINT32_MAX;

	dim3 grid((threads + threadsperblock-1)/threadsperblock);
	dim3 block(threadsperblock);

	/* Check error on Ctrl+C or kill to prevent segfaults on exit */
	if (cudaMemset(d_resNonce[thr_id], 0xff, NBN*sizeof(uint32_t)) != cudaSuccess)
		return result;

	blake256_gpu_hash_16 <<<grid, block>>> (threads, startNonce, d_resNonce[thr_id], highTarget, (int) rounds, opt_tracegpu);
	cudaDeviceSynchronize();
	if (cudaSuccess == cudaMemcpy(h_resNonce[thr_id], d_resNonce[thr_id], NBN*sizeof(uint32_t), cudaMemcpyDeviceToHost)) {
		//cudaThreadSynchronize(); /* seems no more required */
		result = h_resNonce[thr_id][0];
		for (int n=0; n < (NBN-1); n++)
			extra_results[n] = h_resNonce[thr_id][n+1];
	}
	return result;
}

__host__
static void blake256mid(uint32_t *output, const uint32_t *input, int8_t rounds = 14)
{
	sph_blake256_context ctx;

	/* in sph_blake.c */
	blake256_rounds = rounds;

	sph_blake256_init(&ctx);
	sph_blake256(&ctx, input, 64);

	memcpy(output, (void*)ctx.H, 32);
}

__host__
void blake256_cpu_setBlock_16(uint32_t *penddata, const uint32_t *midstate, const uint32_t *ptarget)
{
	uint32_t _ALIGN(64) data[11];
	memcpy(data, midstate, 32);
	data[8] = penddata[0];
	data[9] = penddata[1];
	data[10]= penddata[2];
	CUDA_SAFE_CALL(cudaMemcpyToSymbol(d_data, data, 32 + 12, 0, cudaMemcpyHostToDevice));
}
#endif

extern "C" int scanhash_blake256(int thr_id, uint32_t *pdata, const uint32_t *ptarget,
	uint32_t max_nonce, unsigned long *hashes_done, int8_t blakerounds=14)
{
	const uint32_t first_nonce = pdata[19];
	static bool init[8] = { 0, 0, 0, 0, 0, 0, 0, 0 };
	uint64_t targetHigh = ((uint64_t*)ptarget)[3];
	uint32_t _ALIGN(64) endiandata[20];
#if PRECALC64
	uint32_t _ALIGN(64) midstate[8];
#else
	uint32_t crcsum;
#endif
	int intensity = (device_sm[device_map[thr_id]] > 500) ? 22 : 20;
	uint32_t throughput = opt_work_size ? opt_work_size : (1 << intensity);
	throughput = min(throughput, max_nonce - first_nonce);

	int rc = 0;

#if NBN > 1
	if (extra_results[0] != UINT32_MAX) {
		// possible extra result found in previous call
		if (first_nonce <= extra_results[0] && max_nonce >= extra_results[0]) {
			pdata[19] = extra_results[0];
			*hashes_done = pdata[19] - first_nonce + 1;
			extra_results[0] = UINT32_MAX;
			rc = 1;
			goto exit_scan;
		}
	}
#endif

	if (opt_benchmark) {
		targetHigh = 0x1ULL << 32;
		((uint32_t*)ptarget)[6] = swab32(0xff);
	}

	if (opt_tracegpu) {
		/* test call from util.c */
		throughput = 1;
		for (int k = 0; k < 20; k++)
			pdata[k] = swab32(pdata[k]);
	}

	if (!init[thr_id]) {
		if (num_processors > 1)
			cudaSetDevice(device_map[thr_id]);
		CUDA_CALL_OR_RET_X(cudaMallocHost(&h_resNonce[thr_id], NBN * sizeof(uint32_t)), 0);
		CUDA_CALL_OR_RET_X(cudaMalloc(&d_resNonce[thr_id], NBN * sizeof(uint32_t)), 0);
		init[thr_id] = true;
	}

#if PRECALC64
	for (int k = 0; k < 16; k++)
		be32enc(&endiandata[k], pdata[k]);
	blake256mid(midstate, endiandata, blakerounds);
	blake256_cpu_setBlock_16(&pdata[16], midstate, ptarget);
#else
	blake256_cpu_setBlock_80(pdata, ptarget);
	crcsum = crc32_u32t(pdata, 64);
#endif /* PRECALC64 */

	do {
		uint32_t foundNonce =
#if PRECALC64
		// GPU HASH (second block only, first is midstate)
		blake256_cpu_hash_16(thr_id, throughput, pdata[19], targetHigh, blakerounds);
#else
		// GPU FULL HASH
		blake256_cpu_hash_80(thr_id, throughput, pdata[19], targetHigh, crcsum, blakerounds);
#endif
		if (foundNonce != UINT32_MAX)
		{
			uint32_t vhashcpu[8];
			uint32_t Htarg = (uint32_t)targetHigh;

			for (int k=0; k < 19; k++)
				be32enc(&endiandata[k], pdata[k]);

			be32enc(&endiandata[19], foundNonce);
			blake256hash(vhashcpu, endiandata, blakerounds);

			//applog(LOG_BLUE, "%08x %16llx", vhashcpu[6], targetHigh);
			if (vhashcpu[6] <= Htarg /*&& fulltest(vhashcpu, ptarget)*/)
			{
				pdata[19] = foundNonce;
				rc = 1;

				if (extra_results[0] != UINT32_MAX) {
					// Rare but possible if the throughput is big
					be32enc(&endiandata[19], extra_results[0]);

					blake256hash(vhashcpu, endiandata, blakerounds);
					if (vhashcpu[6] <= Htarg /* && fulltest(vhashcpu, ptarget) */) {
						applog(LOG_NOTICE, "GPU found more than one result " CL_GRN "yippee!");
						rc = 2;
					} else {
						extra_results[0] = UINT32_MAX;
					}
				}

				//applog_hash((uint8_t*)ptarget);
				//applog_compare_hash((uint8_t*)vhashcpu,(uint8_t*)ptarget);
				goto exit_scan;
			}
			else if (opt_debug) {
				applog_hash((uchar*)ptarget);
				applog_compare_hash((uchar*)vhashcpu, (uchar*)ptarget);
				applog(LOG_DEBUG, "GPU #%d: result for nonce %08x does not validate on CPU!", thr_id, foundNonce);
			}
		}

		if ((uint64_t) pdata[19] + throughput > (uint64_t) max_nonce) {
			pdata[19] = max_nonce;
			break;
		}

		pdata[19] += throughput;

	} while (!work_restart[thr_id].restart);

exit_scan:
	*hashes_done = pdata[19] - first_nonce + 1; // (+1 to prevent locks)
	return rc;
}