Taking a tiny part of a huge matrix still matters to calculaltion time ?

Say I have a matrix 'A' sized 10000 x 1000

But I just use A(1:10, 1:10) for a certain purpose.

Of course, it takes shorter time than A(:, :)

But, is it slower significatly than when use matrix B sized 10 x 10 ?

    size(A)
    ans = 10000    10000
    tic
    some_calculation( A(1:10, 1:10) );
    toc

    size(B)
    ans = 10    10
    tic
    some_calculation( B(1:10, 1:10) );
    toc

Answer 1

[per isakson]

Because the sub-matrix is not in a contiguous piece of memory.