no suitable conversion function from "std::vector<unsigned int, std::allocator<unsigned i

Accelerated Computing CUDA CUDA Programming and Performance
1

Hello , I saw in a piece of code this:

std::vector< unsigned int > myarray( mysize );

  std::fill( myarray.begin(), myarray.end(), 0 );

And when I tried to do:

unsigned int   * devmyarray;
cudaMalloc( (void**) &devmyarray, mysize sizeof(unsigned int) )
cudaMemcpy( devmyarray , myarray , mysize sizeof(unsigned int), cudaMemcpyHostToDevice )

it throws:

no suitable conversion function from "std::vector<unsigned int, std::allocator<unsigned int>>" to "const void *" exists
2

Unlike ordinary C arrays, the name of a std::vector is not a pointer to its first element.

instead of this:

cudaMemcpy( devmyarray , myarray , mysize sizeof(unsigned int), cudaMemcpyHostToDevice )

(note that I’m just reproducing the junk code you put in your question above. This line won’t compile.)

try this:

cudaMemcpy( devmyarray ,&(myarray[0]) , mysize*sizeof(unsigned int), cudaMemcpyHostToDevice );

or this:

cudaMemcpy( devmyarray , myarray.data() , mysize*sizeof(unsigned int), cudaMemcpyHostToDevice );

This has nothing to do with CUDA, by the way.

3

Ok , thank you very much for your help txbob.
( I am not used at all to vectors ,I didn’t know)