GLMapBufferObject & copying from VBO to GPU's other part of memory?

Accelerated Computing CUDA CUDA Programming and Performance
1

I have sth like:

//in class:

glGenBuffers(1, &cVBO_vertices);

glBindBuffer(GL_ARRAY_BUFFER, cVBO_vertices);

glBufferData(GL_ARRAY_BUFFER, cVBO_numberVertices*3*sizeof(GLfloat), (GLfloat*)points, GL_DYNAMIC_DRAW);

//*************************

cudaGLRegisterBufferObject(myVBO.cVBO_vertices);

cudaMalloc((void **) &dataGPU, size);

cudaGLMapBufferObject((void**)&dataGPU, myVBO.cVBO_vertices);

//CUDA calculations

//...

cudaGLUnmapBufferObject(myVBO.cVBO_vertices);

I do calculations using dataGPU (not myVBO.cVBO_vertices). Is the data copying from VBO to allocated memory? Is possible to do it directly using VBO?

2

You don’t need to call cudaMalloc, the memory has already been allocated in the VBO. cudaGLMapBufferObject just makes sure the VBO is in GPU memory and returns a GPU pointer to it, there is no copy.

void *dataGPU;

cudaGLMapBufferObject((void**)&dataGPU, myVBO.cVBO_vertices);