CODEWITHSUNDEEP
c++cuda
Sarah Tattersall
Sarah Tattersall

Reputation: 1305

cudaMallocHost is failing to allocate even the smallest of memory

I'm a bit of a CUDA newbie so was wondering if someone could help me out.

I read that pinning can seriously improve your programs performance and so am trying to do exactly that. I'm running my code on a GeForce GT 330 which has compute capability of 1.0.

When I run my program I get that cudaMallocHost failed to allocate memory and so I've condensed my problem down into a small example that can be seen below.

Mesh.hpp

#ifndef MESH_HPP_

#define MESH_HPP_


#include <cstddef>
#include <vector>

#include <driver_types.h>

class Mesh{
public:
  Mesh();
  ~Mesh();  
  void pin_data();

  std::vector<size_t> _a;
  size_t* _a_pinned;

private:
  void cuda_check(cudaError_t success);
};

#endif /* MESH_HPP_ */

Mesh.cpp

#include <iostream>
#include <cmath>
#include <vector>
#include <string.h>

#include <cuda.h>
#include <cuda_runtime.h>

#include "Mesh.hpp"

Mesh::Mesh(){
  for(size_t i = 0; i < 10; i++){
    _a.push_back(i);
  }
}

Mesh::~Mesh() {
  cudaFreeHost(_a_pinned);
}

void Mesh::pin_data() {
  size_t _a_bytes = sizeof(size_t) * _a.size();

  cuda_check(cudaMallocHost((void **)_a_pinned, _a_bytes));
  memcpy(_a_pinned, &_a[0], _a_bytes);
}

void Mesh::cuda_check(cudaError_t status) {
  if (status != cudaSuccess) {
    std::cout << "Error could not allocate memory result " << status << std::endl;
    exit(1);
  }
}

Main.cpp

#include <cstdlib>
#include <iostream>

#include "Mesh.hpp"


int main(int argc, char **argv){

  Mesh *mesh = new Mesh();
  mesh->pin_data();

  delete mesh;

  return EXIT_SUCCESS;
}

When I run my code the output is:

'Error could not allocate memory result 11'

Upvotes: 1

Views: 2865

Answers (1)

Robert Crovella
Robert Crovella

Reputation: 151799

change this line:

cuda_check(cudaMallocHost((void **)_a_pinned, _a_bytes));

to this:

cuda_check(cudaMallocHost((void **)&_a_pinned, _a_bytes));

(only change is to add the ampersand)

cudaMalloc operations expect to modify a pointer value, therefore they must be passed the address of the pointer to modify, not the pointer itself.

That fixed it for me. I'm still a little puzzled by vectors of <size_t> but to each his or her own.

If you want, as a suggestion, in your Mesh:cuda_check method, you might add a line like so:

  std::cout << "Error could not allocate memory result " << status << std::endl;
  std::cout << "Error is: " << cudaGetErrorString(status) << std::endl; //add this line

Upvotes: 5

Related Questions