How are 2D static arrays allocated?

[Saaa]

erm, so what would be the best way to get a large array?
which is also fast.

>>> You can have a dynamic array that _points_ to the stack, but there is
>>> no general mechanism for allocating a dynamic array directly on the
>>> stack.  You might be able to do something with alloca, but stack space
>>> is usually limited and you wouldn't be able to have very large arrays.
>>>
>>> Performance of dynamic arrays is the same no matter where their data
>>> is.  Fixed-size 2D arrays are not faster _because_ they are on the
>>> stack, they just happen to be allocated on the stack.  They are faster
>>> (usually) because they don't need two pointer dereferences.  You can
>>> allocated a fixed-size 2D array on the heap (well.. inside a struct or
>>> class anyway) and it will be just as fast.
>>
>> Ok, so when I want a multimillion array with speed I should use a static
>> array on the heap. That would be using new, right :)
>
> You.. can't actually allocate a static array on the heap using new.
> At least not directly.  It's kind of an embarrassing hole in the
> syntax.  In fact, I'm not sure if you can even get a static array to
> point onto the heap, since a static array "reference" is treated like
> a value type when assigned to, so if you do something like:
>
> int[3][4] b = (new int[3][4][](1))[0];
>
> The weirdness on the right-hand-side is needed to get around the
> compiler "helpfully" rewriting "new int[3][4]" as "new int[][](3, 4)".
> But this code will simply allocate a static array on the heap, and
> then copy its contents onto the stack.  Dumb.
>
> Other than using a struct/class that contains a static array and
> allocating _that_ on the heap, I don't know of any way of getting it
> on the heap.  But then you'd be double-dereferencing anyway since
> you'd have to go through the struct pointer or class reference!  Sigh.