What are the benefits of row major order over column major order when processing 2D arrays?

Question

People prefer accessing a 2D array in row-major order than column-major order. Why row-major is efficient? Is there some advantage of row major as compare to column major wrt memory.

Answer 1

The benefits are purely related to performance.

If your array is stored in a row-major format and you are accessing elements row-wise, the cache of your processor will have the next elements already loaded when you need them. And after a few accesses, the pre-fetcher will kick in and start bringing in the next elements in advance. If you access elements in column-major format, you will be using only one element from a cache line, and 'wasting' the fetched elements.

Note the difference between C (and C++) and Fortran: the latter stores arrays in column-major format.

In high-performance computing applications, the code very rarely explicitly declares a multi-dimensional array. We typically declare a 1D array and translate the multiple dimensions into the right index in the code. This can lead to many optimizations, depending on what you're trying to do!