I have a voxel structure as:
struct voxel
{
unsigned char R, G, B;
voxel()
{
R = G = B = 0;
}
//parameteric contructor with parameters
voxel(unsigned char pR, unsigned char pG, unsigned char pB)
{
R = pR; G = pG; B = pB;
}
};
I have a very large n for number of voxels.
int n = 300 * 300 * 300;
Now when i initialize voxels with vector it takes approximately 79 MB in RAM.
std::vector< voxel > vi(n);
But it takes more than 2 GB when i initialize it this way using shared_ptr and stack overflows.
std::vector< std::shared_ptr<voxel> > vi(n);
for (size_t i = 0; i < n; i++)
{
vi.push_back(std::shared_ptr<voxel>(new voxel()));
}
What could be the reason for this behavior and how can i avoid it?
Additional Notes:
std::vector< std::shared_ptr<voxel> > vi(n); //statement takes 211 MB alone
Update: I have also tried with this loop instead of push back but the result is same. I now have a general gist of why is this happening.
for (size_t i = 0; i < n; i++)
{
vi[i].reset(new voxel());
vi[i]->B = 0;
vi[i]->R = 0;
vi[i]->G = 0;
}
std::vector< voxel > vi(n);
Is going to take up sizeof(voxel) * n
bytes of memory. When you change to a shared pointer you are now going to have the cost of the shared pointer and the the voxel
. That would be equivilent to
sizeof(voxel) * n + sizeof(std::shared_ptr<voxel>) * n
Where the sizeof(std::shared_ptr<voxel>)
is likely to be 16 bytes.
You are also wasting a lot of space in your second example as you declare
std::vector< std::shared_ptr<voxel> > vi(n);
Which is going to create n
empty shared_ptr
s and then you push_back
another n
non empty shared_ptr
s so you doulbe the size of the vector. If you want to preallocate the size of the vector
then you should use
std::vector< std::shared_ptr<voxel> > vi;
vi.reserve(n);