ccminer/qubit/qubit_luffa512.cu

/*
 * luffa_for_32.c
 * Version 2.0 (Sep 15th 2009)
 *
 * Copyright (C) 2008-2009 Hitachi, Ltd. All rights reserved.
 *
 * Hitachi, Ltd. is the owner of this software and hereby grant
 * the U.S. Government and any interested party the right to use
 * this software for the purposes of the SHA-3 evaluation process,
 * notwithstanding that this software is copyrighted.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */
#include <stdio.h>
#include <stdint.h>
#include <memory.h>

#include "cuda_helper.h"

#ifndef UINT32_MAX
#define UINT32_MAX UINT_MAX
#endif

typedef unsigned char BitSequence;

__constant__ uint64_t c_PaddedMessage80[16]; // padded message (80 bytes + padding)
__constant__ uint32_t c_Target[8];

static uint32_t *h_resNounce[MAX_GPUS];
static uint32_t *d_resNounce[MAX_GPUS];

#define NBN 1 /* max results, could be 2, see blake32.cu */
#if NBN > 1
static uint32_t extra_results[2] = { UINT32_MAX, UINT32_MAX };
#endif

typedef struct {
	uint32_t buffer[8]; /* Buffer to be hashed */
	uint32_t chainv[40];   /* Chaining values */
} hashState;

#define BYTES_SWAP32(x) cuda_swab32(x)

#define MULT2(a,j)\
	tmp = a[7+(8*j)];\
	a[7+(8*j)] = a[6+(8*j)];\
	a[6+(8*j)] = a[5+(8*j)];\
	a[5+(8*j)] = a[4+(8*j)];\
	a[4+(8*j)] = a[3+(8*j)] ^ tmp;\
	a[3+(8*j)] = a[2+(8*j)] ^ tmp;\
	a[2+(8*j)] = a[1+(8*j)];\
	a[1+(8*j)] = a[0+(8*j)] ^ tmp;\
	a[0+(8*j)] = tmp;

#define TWEAK(a0,a1,a2,a3,j)\
	a0 = (a0<<(j))|(a0>>(32-j));\
	a1 = (a1<<(j))|(a1>>(32-j));\
	a2 = (a2<<(j))|(a2>>(32-j));\
	a3 = (a3<<(j))|(a3>>(32-j));

#define STEP(c0,c1)\
	SUBCRUMB(chainv[0],chainv[1],chainv[2],chainv[3],tmp);\
	SUBCRUMB(chainv[5],chainv[6],chainv[7],chainv[4],tmp);\
	MIXWORD(chainv[0],chainv[4]);\
	MIXWORD(chainv[1],chainv[5]);\
	MIXWORD(chainv[2],chainv[6]);\
	MIXWORD(chainv[3],chainv[7]);\
	ADD_CONSTANT(chainv[0],chainv[4],c0,c1);

#define SUBCRUMB(a0,a1,a2,a3,a4)\
	a4  = a0;\
	a0 |= a1;\
	a2 ^= a3;\
	a1  = ~a1;\
	a0 ^= a3;\
	a3 &= a4;\
	a1 ^= a3;\
	a3 ^= a2;\
	a2 &= a0;\
	a0  = ~a0;\
	a2 ^= a1;\
	a1 |= a3;\
	a4 ^= a1;\
	a3 ^= a2;\
	a2 &= a1;\
	a1 ^= a0;\
	a0  = a4;

#define MIXWORD(a0,a4)\
	a4 ^= a0;\
	a0  = (a0<<2) | (a0>>(30));\
	a0 ^= a4;\
	a4  = (a4<<14) | (a4>>(18));\
	a4 ^= a0;\
	a0  = (a0<<10) | (a0>>(22));\
	a0 ^= a4;\
	a4  = (a4<<1) | (a4>>(31));

#define ADD_CONSTANT(a0,b0,c0,c1)\
	a0 ^= c0;\
	b0 ^= c1;

/* initial values of chaining variables */
__constant__ uint32_t c_IV[40];
const uint32_t h2_IV[40] = {
	0x6d251e69,0x44b051e0,0x4eaa6fb4,0xdbf78465,
	0x6e292011,0x90152df4,0xee058139,0xdef610bb,
	0xc3b44b95,0xd9d2f256,0x70eee9a0,0xde099fa3,
	0x5d9b0557,0x8fc944b3,0xcf1ccf0e,0x746cd581,
	0xf7efc89d,0x5dba5781,0x04016ce5,0xad659c05,
	0x0306194f,0x666d1836,0x24aa230a,0x8b264ae7,
	0x858075d5,0x36d79cce,0xe571f7d7,0x204b1f67,
	0x35870c6a,0x57e9e923,0x14bcb808,0x7cde72ce,
	0x6c68e9be,0x5ec41e22,0xc825b7c7,0xaffb4363,
	0xf5df3999,0x0fc688f1,0xb07224cc,0x03e86cea};

__constant__ uint32_t c_CNS[80];
uint32_t h2_CNS[80] = {
	0x303994a6,0xe0337818,0xc0e65299,0x441ba90d,
	0x6cc33a12,0x7f34d442,0xdc56983e,0x9389217f,
	0x1e00108f,0xe5a8bce6,0x7800423d,0x5274baf4,
	0x8f5b7882,0x26889ba7,0x96e1db12,0x9a226e9d,
	0xb6de10ed,0x01685f3d,0x70f47aae,0x05a17cf4,
	0x0707a3d4,0xbd09caca,0x1c1e8f51,0xf4272b28,
	0x707a3d45,0x144ae5cc,0xaeb28562,0xfaa7ae2b,
	0xbaca1589,0x2e48f1c1,0x40a46f3e,0xb923c704,
	0xfc20d9d2,0xe25e72c1,0x34552e25,0xe623bb72,
	0x7ad8818f,0x5c58a4a4,0x8438764a,0x1e38e2e7,
	0xbb6de032,0x78e38b9d,0xedb780c8,0x27586719,
	0xd9847356,0x36eda57f,0xa2c78434,0x703aace7,
	0xb213afa5,0xe028c9bf,0xc84ebe95,0x44756f91,
	0x4e608a22,0x7e8fce32,0x56d858fe,0x956548be,
	0x343b138f,0xfe191be2,0xd0ec4e3d,0x3cb226e5,
	0x2ceb4882,0x5944a28e,0xb3ad2208,0xa1c4c355,
	0xf0d2e9e3,0x5090d577,0xac11d7fa,0x2d1925ab,
	0x1bcb66f2,0xb46496ac,0x6f2d9bc9,0xd1925ab0,
	0x78602649,0x29131ab6,0x8edae952,0x0fc053c3,
	0x3b6ba548,0x3f014f0c,0xedae9520,0xfc053c31};


/***************************************************/
__device__ __forceinline__
void rnd512(hashState *state)
{
	int i,j;
	uint32_t t[40];
	uint32_t chainv[8];
	uint32_t tmp;

#pragma unroll 8
	for(i=0;i<8;i++) {
		t[i]=0;
#pragma unroll 5
		for(j=0;j<5;j++) {
			t[i] ^= state->chainv[i+8*j];
		}
	}

	MULT2(t, 0);

#pragma unroll 5
	for(j=0;j<5;j++) {
#pragma unroll 8
		for(i=0;i<8;i++) {
			state->chainv[i+8*j] ^= t[i];
		}
	}

#pragma unroll 5
	for(j=0;j<5;j++) {
#pragma unroll 8
		for(i=0;i<8;i++) {
			t[i+8*j] = state->chainv[i+8*j];
		}
	}

#pragma unroll 5
	for(j=0;j<5;j++) {
		MULT2(state->chainv, j);
	}

#pragma unroll 5
	for(j=0;j<5;j++) {
#pragma unroll 8
		for(i=0;i<8;i++) {
			state->chainv[8*j+i] ^= t[8*((j+1)%5)+i];
		}
	}

#pragma unroll 5
	for(j=0;j<5;j++) {
#pragma unroll 8
		for(i=0;i<8;i++) {
			t[i+8*j] = state->chainv[i+8*j];
		}
	}

#pragma unroll 5
	for(j=0;j<5;j++) {
		MULT2(state->chainv, j);
	}

#pragma unroll 5
	for(j=0;j<5;j++) {
#pragma unroll 8
		for(i=0;i<8;i++) {
			state->chainv[8*j+i] ^= t[8*((j+4)%5)+i];
		}
	}

#pragma unroll 5
	for(j=0;j<5;j++) {
#pragma unroll 8
		for(i=0;i<8;i++) {
			state->chainv[i+8*j] ^= state->buffer[i];
		}
		MULT2(state->buffer, 0);
	}

#pragma unroll 8
	for(i=0;i<8;i++) {
		chainv[i] = state->chainv[i];
	}

#pragma unroll 8
	for(i=0;i<8;i++) {
		STEP(c_CNS[(2*i)],c_CNS[(2*i)+1]);
	}

#pragma unroll 8
	for(i=0;i<8;i++) {
		state->chainv[i] = chainv[i];
		chainv[i] = state->chainv[i+8];
	}

	TWEAK(chainv[4],chainv[5],chainv[6],chainv[7],1);

#pragma unroll 8
	for(i=0;i<8;i++) {
		STEP(c_CNS[(2*i)+16],c_CNS[(2*i)+16+1]);
	}

#pragma unroll 8
	for(i=0;i<8;i++) {
		state->chainv[i+8] = chainv[i];
		chainv[i] = state->chainv[i+16];
	}

	TWEAK(chainv[4],chainv[5],chainv[6],chainv[7],2);

#pragma unroll 8
	for(i=0;i<8;i++) {
		STEP(c_CNS[(2*i)+32],c_CNS[(2*i)+32+1]);
	}

#pragma unroll 8
	for(i=0;i<8;i++) {
		state->chainv[i+16] = chainv[i];
		chainv[i] = state->chainv[i+24];
	}

	TWEAK(chainv[4],chainv[5],chainv[6],chainv[7],3);

#pragma unroll 8
	for(i=0;i<8;i++) {
		STEP(c_CNS[(2*i)+48],c_CNS[(2*i)+48+1]);
	}

#pragma unroll 8
	for(i=0;i<8;i++) {
		state->chainv[i+24] = chainv[i];
		chainv[i] = state->chainv[i+32];
	}

	TWEAK(chainv[4],chainv[5],chainv[6],chainv[7],4);

#pragma unroll 8
	for(i=0;i<8;i++) {
		STEP(c_CNS[(2*i)+64],c_CNS[(2*i)+64+1]);
	}

#pragma unroll 8
	for(i=0;i<8;i++) {
		state->chainv[i+32] = chainv[i];
	}
}


__device__ __forceinline__
void Update512(hashState *state, const BitSequence *data)
{
#pragma unroll 8
	for(int i=0;i<8;i++) state->buffer[i] = BYTES_SWAP32(((uint32_t*)data)[i]);
	rnd512(state);

#pragma unroll 8
	for(int i=0;i<8;i++) state->buffer[i] = BYTES_SWAP32(((uint32_t*)(data+32))[i]);
	rnd512(state);
#pragma unroll 4
	for(int i=0;i<4;i++) state->buffer[i] = BYTES_SWAP32(((uint32_t*)(data+64))[i]);
}


/***************************************************/
__device__ __forceinline__
void finalization512(hashState *state, uint32_t *b)
{
	int i,j;

	state->buffer[4] = 0x80000000;
#pragma unroll 3
	for(int i=5;i<8;i++) state->buffer[i] = 0;
	rnd512(state);

	/*---- blank round with m=0 ----*/
#pragma unroll 8
	for(i=0;i<8;i++) state->buffer[i] =0;
	rnd512(state);

#pragma unroll 8
	for(i=0;i<8;i++) {
		b[i] = 0;
#pragma unroll 5
		for(j=0;j<5;j++) {
			b[i] ^= state->chainv[i+8*j];
		}
		b[i] = BYTES_SWAP32((b[i]));
	}

#pragma unroll 8
	for(i=0;i<8;i++) state->buffer[i]=0;
	rnd512(state);

#pragma unroll 8
	for(i=0;i<8;i++) {
		b[8+i] = 0;
#pragma unroll 5
		for(j=0;j<5;j++) {
			b[8+i] ^= state->chainv[i+8*j];
		}
		b[8+i] = BYTES_SWAP32((b[8+i]));
	}
}


/***************************************************/
// Die Hash-Funktion
__global__
void qubit_luffa512_gpu_hash_80(int threads, uint32_t startNounce, void *outputHash)
{
	int thread = (blockDim.x * blockIdx.x + threadIdx.x);
	if (thread < threads)
	{
		uint32_t nounce = startNounce + thread;
		union {
		uint64_t buf64[16];
		uint32_t buf32[32];
		} buff;

#pragma unroll 16
		for (int i=0; i < 16; ++i) buff.buf64[i] = c_PaddedMessage80[i];

		// die Nounce durch die thread-spezifische ersetzen
		buff.buf64[9] = REPLACE_HIWORD(buff.buf64[9], cuda_swab32(nounce));


		hashState state;
#pragma unroll 40
		for(int i=0;i<40;i++) state.chainv[i] = c_IV[i];
#pragma unroll 8
		for(int i=0;i<8;i++) state.buffer[i] = 0;
		Update512(&state, (BitSequence*)buff.buf32);
		uint32_t *outHash = (uint32_t *)outputHash + 16 * thread;
		finalization512(&state, (uint32_t*)outHash);
	}
}

__global__
void qubit_luffa512_gpu_finalhash_80(int threads, uint32_t startNounce, void *outputHash, uint32_t *resNounce)
{
	int thread = (blockDim.x * blockIdx.x + threadIdx.x);
	if (thread < threads)
	{
		uint32_t nounce = startNounce + thread;
		union {
			uint64_t buf64[16];
			uint32_t buf32[32];
		} buff;
		uint32_t Hash[16];

		#pragma unroll 16
		for (int i=0; i < 16; ++i) buff.buf64[i] = c_PaddedMessage80[i];

		// Tested nonce
		buff.buf64[9] = REPLACE_HIWORD(buff.buf64[9], cuda_swab32(nounce));

		hashState state;
		#pragma unroll 40
		for(int i=0;i<40;i++) state.chainv[i] = c_IV[i];

		#pragma unroll 8
		for(int i=0;i<8;i++) state.buffer[i] = 0;

		Update512(&state, (BitSequence*)buff.buf32);
		finalization512(&state, Hash);

		/* dont ask me why not a simple if (Hash[i] > c_Target[i]) return;
		 * we lose 20% in perfs without the position test */
		int position = -1;
		#pragma unroll 8
		for (int i = 7; i >= 0; i--) {
			if (Hash[i] > c_Target[i]) {
				if (position < i) {
					return;
				}
			}
			if (Hash[i] < c_Target[i]) {
				if (position < i) {
					position = i;
					//break; /* impact perfs, unroll ? */
				}
			}
		}

#if NBN == 1
		if (resNounce[0] > nounce) {
			resNounce[0] = nounce;
		}
#else
		/* keep the smallest nounce, + extra one if found */
		if (resNounce[0] > nounce) {
			resNounce[1] = resNounce[0];
			resNounce[0] = nounce;
		} else {
			resNounce[1] = nounce;
		}
#endif
	}
}

__host__
void qubit_luffa512_cpu_init(int thr_id, int threads)
{
	CUDA_SAFE_CALL(cudaMemcpyToSymbol(c_IV, h2_IV, sizeof(h2_IV), 0, cudaMemcpyHostToDevice));
	CUDA_SAFE_CALL(cudaMemcpyToSymbol(c_CNS, h2_CNS, sizeof(h2_CNS), 0, cudaMemcpyHostToDevice));
	CUDA_SAFE_CALL(cudaMalloc(&d_resNounce[thr_id], NBN * sizeof(uint32_t)));
	CUDA_SAFE_CALL(cudaMallocHost(&h_resNounce[thr_id], NBN * sizeof(uint32_t)));
}

__host__
uint32_t qubit_luffa512_cpu_finalhash_80(int thr_id, int threads, uint32_t startNounce, uint32_t *d_outputHash,int order)
{
	uint32_t result = UINT32_MAX;
	cudaMemset(d_resNounce[thr_id], 0xff, NBN * sizeof(uint32_t));
	const int threadsperblock = 256;

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

	size_t shared_size = 0;

	qubit_luffa512_gpu_finalhash_80 <<<grid, block, shared_size>>> (threads, startNounce, d_outputHash, d_resNounce[thr_id]);
	cudaThreadSynchronize();
	if (cudaSuccess == cudaMemcpy(h_resNounce[thr_id], d_resNounce[thr_id], NBN * sizeof(uint32_t), cudaMemcpyDeviceToHost)) {
		//cudaThreadSynchronize();
		result = h_resNounce[thr_id][0];
#if NBN > 1
		extra_results[0] = h_resNounce[thr_id][1];
#endif
	}
	return result;
}

__host__
void qubit_luffa512_cpu_hash_80(int thr_id, int threads, uint32_t startNounce, uint32_t *d_outputHash,int order)
{
	const int threadsperblock = 256;

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

	qubit_luffa512_gpu_hash_80 <<<grid, block, shared_size>>> (threads, startNounce, d_outputHash);
	MyStreamSynchronize(NULL, order, thr_id);
}

__host__
void qubit_luffa512_cpu_setBlock_80(void *pdata)
{
	unsigned char PaddedMessage[128];

	memcpy(PaddedMessage, pdata, 80);
	memset(PaddedMessage+80, 0, 48);
	PaddedMessage[80] = 0x80;
	PaddedMessage[111] = 1;
	PaddedMessage[126] = 0x02;
	PaddedMessage[127] = 0x80;

	CUDA_SAFE_CALL(cudaMemcpyToSymbol( c_PaddedMessage80, PaddedMessage, 16*sizeof(uint64_t), 0, cudaMemcpyHostToDevice));
}

__host__
void qubit_luffa512_cpufinal_setBlock_80(void *pdata, const void *ptarget)
{
	unsigned char PaddedMessage[128];

	memcpy(PaddedMessage, pdata, 80);
	memset(PaddedMessage+80, 0, 48);
	PaddedMessage[80] = 0x80;
	PaddedMessage[111] = 1;
	PaddedMessage[126] = 0x02;
	PaddedMessage[127] = 0x80;

	CUDA_SAFE_CALL(cudaMemcpyToSymbol(c_Target, ptarget, 8*sizeof(uint32_t), 0, cudaMemcpyHostToDevice));
	CUDA_SAFE_CALL(cudaMemcpyToSymbol(c_PaddedMessage80, PaddedMessage, 16*sizeof(uint64_t), 0, cudaMemcpyHostToDevice));
}