I had to develop a 3DArray object, but i wondered what was the best model for cpu access.

I wrote 2 models : array[a][b][c] and array[a].content[b][c]

I thought that the first model would be much difficult for the cpu to access and it would prefer the second form.

And i was wrong :), the first form is reaaaallly faster as the benchmark proves it :

Hope this benchmark will help some of you...