alphaville

/* FILE: cuda_timer.cuh */

#include <cuda_runtime.h>
#include "error_handles.cuh"

static cudaEvent_t start;
static cudaEvent_t stop;
static short timer_running = 0;
static short tic_called = 0;

alphaville

#include <stdio.h>
#include <stdlib.h>
#include <cuda_runtime.h>
#include <assert.h>
#include "helper_cuda.h"
#include "rand_data.h"
#include <math.h>

#define 	BLOCK_SIZE 		16

alphaville

template<typename T>
__global__ void matvec_kernel(
  const T * dA,
  const T * dx,
  T * dy,
  const unsigned int * nRows,
  const unsigned int * nx)
{

  unsigned int bid = blockIdx.x;

alphaville

/* Auto-generated file (by a MATLAB script) - ID 3062321795 */

#ifndef __RAND_DATA_TURBO__
#define __RAND_DATA_TURBO__

#define NS 32

__device__ float dev_x[32] = 
 { 9.320295340189, 8.666071467758, 2.064062819139, 5.206786781683, 9.887525026886, 9.196139126590, 2.988093813008, 9.115010659100, 4.994307851747, 0.409423900211, 
   1.911173066863, 1.515824153316, 5.797111859887, 1.448878331722, 8.210170746106, 6.347733121928, 7.880805428768, 0.672098319650, 5.452277295227, 9.969671220601, 

alphaville

int main(void) {

  checkCudaErrors(cudaSetDeviceFlags(cudaDeviceMapHost));

  float * dev_ptr_A = NULL;
  float * dev_ptr_x = NULL;
  float * hst_y = NULL;
  float * address_y = NULL;

  checkCudaErrors(

alphaville

#include <stdio.h>
#include <stdlib.h>
#include <cuda_runtime.h>
#include "cublas_v2.h"
#include "helper_cuda.h"

#define ns 6

__device__ float dev_static_x[6] = {0.0, 1.0, 2.0, 3.0, 4.0, 5.0};
__device__ float dev_static_y[6] = {0.0, -1.0, -2.0, -3.0, -4.0, -5.0};

alphaville

#include <stdio.h>
#include <stdlib.h>
#include <cuda_runtime.h>
#include "cublas_v2.h"
#include "helper_cuda.h"

#define ns 1e8


int main(void) {

alphaville

cudaDeviceProp prop;
 int whichDevice;

 checkCudaErrors( cudaGetDevice(&whichDevice));
 checkCudaErrors( cudaGetDeviceProperties(&prop, whichDevice));

 if (prop.canMapHostMemory != 1){
  fprintf(stderr, "Device cannot map memory!\n");
  return 1;
 }

alphaville

int main(void) {

 int * host_p; /*< Host data allocated as pinned memory */
 int * dev_ptr_p; /*< this pointer resides on the host */
 int ns = 32;
 int data_size = ns * sizeof(int);

 checkCudaErrors(
   cudaHostAlloc((void**) &host_p, data_size, cudaHostAllocMapped));

alphaville

#include <stdio.h>  
#include <cuda_runtime.h>  
#include "helper_cuda.h"  

/* A very simple kernel function */
 __global__ void kernel(int *d_var) { d_var[threadIdx.x] += 10; } 
 
 int * host_p;  
 int * host_result;  
 int * dev_p;