dynamic allocation/deallocation of 2D & 3D arrays

Question

I know about algorithms to allocate/deallocate a 2D array dynamically, however I\'m not too sure about the same for 3D arrays.
Using this knowledge and a bit of symmetry

Answer 1

This is a version of the idea in the question but using only one malloc, inspired by the other answers. It allows intuitive use of the square brackets and easy cleaning. I hope it does not make any compiler implementation specific assumption.

int main(int argc, char *argv[])
{
  int **array, i, j;
  array = allocate2d(3, 4);
  for (i = 0; i < 3; i++)
  {
    for (j = 0; j < 4; j++)
    {
      array[i][j] = j + i + 1;
    }
  }
  for (i = 0; i < 3; i++)
  {
    for (j = 0; j < 4; j++)
    {
      printf("array[%d][%d] = %d\n", i, j, array[i][j]);
    }
  }
  free(array);
  return EXIT_SUCCESS;
}

int **allocate2d(int x, int y)
{
  int i;
  int **array = malloc(sizeof(int *) * x + sizeof(int) * x * y);
  for (i = 0; i < x; i++)
  {
    array[i] = ((int *)(array + x)) + y * i;
  }
  return array;
}

Answer 2

You can see the below code:

#include <stdio.h>
#include <stdlib.h>

void main()
{
    //  Array 3 Dimensions
    int x = 4, y = 5, z = 6;

    //  Array Iterators
    int i, j, k;

    //  Allocate 3D Array
    int *allElements = malloc(x * y * z * sizeof(int));
    int ***array3D = malloc(x * sizeof(int **));

    for(i = 0; i < x; i++)
    {
        array3D[i] = malloc(y * sizeof(int *));

        for(j = 0; j < y; j++)
        {
            array3D[i][j] = allElements + (i * y * z) + (j * z);
        }
    }

    //  Access array elements
    for(i = 0; i < x; i++)
    {
        printf("%d\n", i);

        for(j = 0; j < y; j++)
        {
            printf("\n");

            for(k = 0; k < z; k++)
            {
                array3D[i][j][k] = (i * y * z) + (j * z) + k;
                printf("\t%d", array3D[i][j][k]);
            }
        }

        printf("\n\n");
    }

    //  Deallocate 3D array
    free(allElements);
    for(i = 0; i < x; i++)
    {
        free(array3D[i]);
    }
    free (array3D);
}

For more details see this link 3d array

Answer 3

You can also allocate one array and compute individual indices. This requires fewer allocator calls and results in both less fragmentation and better cache use.

typedef struct {
  int a;
  int b;
  int* data;
} Int2d;

Int2d arr2d = { 2, 3 };
arr2d.data = malloc(arr2d.a * arr2d.b * sizeof *arr2d.data);

Now arr2d[r][c] becomes arr2d.data[r * arr2d.b + c]. Deallocation is a single free() away. As a bonus you're sure to always keep your dynamic array sizes with you.

Extrapolating to 3d:

typedef struct {
  int a;
  int b;
  int c;
  int* data;
} Int3d;

Int3d arr3d = { 2, 3, 4 };
arr3d.data = malloc(arr3d.a * arr3d.b * arr3d.c * sizeof *arr3d.data);

//arr3d[r][c][d]
// becomes:
arr3d.data[r * (arr3d.b * arr3d.c) + c * arr3d.c + d];

You should encapsulate these index operations (and the (de-)allocations for that matter) in a separate function or macro.

(The names for r, c, and d could be better—I was going for row, column, and depth. While a, b, and c are the limits of their corresponding dimensions, you might prefer something like n1, n2, n3 there, or even use an array for them.)

Answer 4

arr3d should be a triple pointer and not just an int. Otherwise looks OK:

void deallocate3D(int*** arr3D,int l,int m)
{
    int i,j;

    for(i=0;i<l;i++)
    {
        for(int j=0;j<m;j++)
        {
                free(arr3D[i][j]);
        }
        free(arr3D[i]);
    }
    free(arr3D);
}

arr3D is a pointer-to-pointer-to-pointer, so arr3D[i] is a pointer-to-pointer and arr3D[i][j] just a pointer. It's correct to release the lowest dimension in a loop first, and then climb up the dimensions until arr3D itself is released.

Also it's more idiomatic to give malloc the sizeof of the pointed type implicitly. Instead of:

  arr3D[i] = (int**)malloc(m * sizeof(int*));

Make it:

  arr3D[i] = (int**)malloc(m * sizeof(*arr3D[i]));

And yes, such dynamically allocated multi-dimensional arrays can be accessed just as statically allocated multi-dimensional arrays.