c++ - Difference in speed computation between STL vector of vectors and contiguous storage -

in program have scroll lot of time 2d array , computation on it; in first way saved matrix vector<vector<int>>;. after tried contiguous memory 2d array

template <typename t> t** create2darray(unsigned nrows, unsigned ncols) {    t** ptr = new t*[nrows];  // allocate pointers    t* pool = new t[nrows*ncols];  // allocate pool    (unsigned = 0; < nrows; ++i, pool += ncols )        ptr[i] = pool;    return ptr; }  template <typename t> void delete2darray(t** arr) {    delete [] arr[0];  // remove pool    delete [] arr;     // remove pointers }  int main() {     double **dptr = create2darray<double>(10,10);    dptr[0][0] = 10;  // example    delete2darray(dptr);  // free memory } 

and saw vector of vectors storaging computation speed more or less 2 time faster contiguous way.i don't understand why.

edit :i have matrix of integer made 0,1,2 , have read file in stored matrix cars (1=bluecar, 2=redcar, 0=empty).

i need write algorithm move cars of matrix in way:

• blue ones move downward;
• red ones move rightward;
• there turn in blue ones move , turn move red ones.

i have 2 vectors have coordinates of non 0 cell of matrix,and scroll matrix through these vectors. have access lot of time matrix , change it.