Consider the following, and correct me if I'm wrong please.

// 10x10 array for this example


#define ARRAYWIDTH 10
#define ARRAYHEIGHT 10
#define I(x,y) (y * ARRAYWIDTH + x)


// First method


// creation

int *2darray = new int[ARRAYWIDTH * ARRAYHEIGHT ];
// indexing

int element = 2darray[I(5,5)];


// Second method


// creation

int **2darray = new int*[ARRAYHEIGHT];
for(int i = 0; i < ARRAYHEIGHT; i++)
	2darray[i] = new int[ARRAYWIDTH];

// indexing

int element = 2darray[5][5];

 

It would seem to me, that method 1 would be more efficient, since you are guaranteed that the 2d array is all together in memory & is more cache friendly than method 2. Is this correct? Also, it's true that each "row" of the array in method 2 could be in completely different areas of memory right? Obviously method 2 would be slower in the allocation phase, so my question is more geared towards accessing the array after creation. Thanks [edited by - DrEvil on April 20, 2004 3:23:30 AM]

quote:
But you get to use 2D array access operators on it instead of some hackified macro/function, so I suppose that''s nice.

I don''t think that IS nice (although macros should be avoided). The longhand "y * ARRAYWIDTH + x" is the nicest way in my opinion.

Mark
Sound Effects For Game Developers
http://www.indiesfx.co.uk
Flatspace is here: The ultimate space adventure
http://www.lostinflatspace.com