utils_cuda.h

Go to the documentation of this file.

/***************************************************************************
 *   Copyright (C) 2014 by PIR (University of Oviedo) and                  *
 *   INCO2 (Polytechnic University of Valencia) groups.                    *
 *   nnmfpack@gmail.com                                                    *
 *                                                                         *
 *   This program is free software; you can redistribute it and/or modify  *
 *   it under the terms of the GNU General Public License as published by  *
 *   the Free Software Foundation; either version 2 of the License, or     *
 *   (at your option) any later version.                                   *
 *                                                                         *
 *   This program is distributed in the hope that it will be useful,       *
 *   but WITHOUT ANY WARRANTY; without even the implied warranty of        *
 *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the         *
 *   GNU General Public License for more details.                          *
 *                                                                         *
 *   You should have received a copy of the GNU General Public License     *
 *   along with this program; if not, write to the                         *
 *   Free Software Foundation, Inc.,                                       *
 *   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.             *
 ***************************************************************************
*/
#ifndef UTILSCUDA_H
#define UTILSCUDA_H

#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <cuda.h>
#include <curand.h>
#include <cuda_runtime.h>
#include <cublas_v2.h>
#include <assert.h>

#define CUDAERR(x) do { if((x)!=cudaSuccess) { \
    printf("CUDA error: %s : %s, line %d\n", cudaGetErrorString(x), __FILE__, __LINE__);\
    return EXIT_FAILURE;}} while(0)

#define CUBLASERR(x) do { if((x)!=CUBLAS_STATUS_SUCCESS) { \
    printf("CUBLAS error: %s, line %d\n", __FILE__, __LINE__);\
    return EXIT_FAILURE;}} while(0)

#define max(a,b) (((a)>(b ))?( a):(b))
#define min(a,b) (((a)<(b ))?( a):(b))

#define fpe(x) (isnan(x) || isinf(x))

/* kernels */
__global__ void vdmemset_cuda(const int n, double *x, const double val);
__global__ void vsmemset_cuda(const int n, float  *x, const float  val);

__global__ void vddiv_cuda(const int n, const double* __restrict__ x, const double* __restrict__ y, double *z);
__global__ void vsdiv_cuda(const int n, const float*  __restrict__ x, const float*  __restrict__ y, float  *z);

__global__ void vdsub_cuda(const int n, const double* __restrict__ x, double *y);
__global__ void vssub_cuda(const int n, const float*  __restrict__ x, float  *y);

__global__ void vderrorbd0_cuda(const int n, const double* __restrict__ x, double *y);
__global__ void vserrorbd0_cuda(const int n, const float*  __restrict__ x, float  *y);

__global__ void vderrorbd1_cuda(const int n, const double* __restrict__ x, double *y);
__global__ void vserrorbd1_cuda(const int n, const float*  __restrict__ x, float  *y);

__global__ void vderrorbdg_cuda(const int n, const double* __restrict__ x, double *y, const double beta);
__global__ void vserrorbdg_cuda(const int n, const float*  __restrict__ x, float  *y, const double beta);


/* wrappers for the kernels */
void dmemset_cuda(const int n, double *x, const double val, cudaStream_t stream);
void smemset_cuda(const int n, float  *x, const float  val, cudaStream_t stream);

void ddiv_cuda(const int n, const double *x, double *y, cudaStream_t stream);
void sdiv_cuda(const int n, const float  *x, float  *y, cudaStream_t stream);

void dsub_cuda(const int n, const double *x, double *y);
void ssub_cuda(const int n, const float  *x, float  *y);

void dlarngenn_cuda(const int m, const int n, const int seed, double *x);
void slarngenn_cuda(const int m, const int n, const int seed, float  *x);

double derror_cuda(const int m, const int n, const int k, const double *x, const double *y, const double *z);
float  serror_cuda(const int m, const int n, const int k, const float  *x, const float  *y, const float  *z);

double derrorbd_cuda(const int m, const int n, const int k, const double *A, const double *W, const double *H, const double beta);
float  serrorbd_cuda(const int m, const int n, const int k, const float  *A, const float  *W, const float  *H, const float  beta);

#endif