Which is best for a list of points: an array of point structs or a 2 dimensional array?

Question

I need an array where each member has 3 elements of the same type. I can use an array of structs, or a 2 dimensional array.

// Array of structs:
struct point {  double x, y, z; };
struct point path[20000];  
// Now access z by
path[i].z

// 2-dimensional array:
double path[20000][3];
// Access z by
path[i][2]

Which method is most efficient? Or are they the same?

Efficiency is really hard to talk about in the abstract, because different approaches, and different trade-offs. In this case, they're about the same, but the named `x`/`y`/`z` fields read way better. Plus, the `Point` abstraction is just really useful overall — Alexander, Sep 17 '20 at 18:11
That's not a two-dimensional array. [What are the differences between a multidimensional array and an array of arrays?](https://stackoverflow.com/questions/597720/what-are-the-differences-between-a-multidimensional-array-and-an-array-of-arrays) — John Wu, Sep 17 '20 at 19:26
I'd venture to suggest that, unless you're trying to save bytes or microseconds, you should use the version that has most expressive power. Someone has to maintain this stuff, after all. — Kevin Boone, Sep 17 '20 at 19:29

score 0 · Answer 1 · answered Sep 17 '20 at 18:11

0

There's no difference.

If I remember correctly, in theory a compiler is allowed to insert padding bytes between the members of struct point, but no sane compiler would do it in this case.

answered Sep 17 '20 at 18:11

HolyBlackCat

78,603
9
131
207

score 0 · Answer 2 · answered Sep 17 '20 at 18:17

struct point { double x, y, z; }; may incur padding - although quite unlikely.
double path[20000][3]; has no padding.

In this case, I recommend to code for clarity.
Since the comment hints that the access is of z, use .z rather than [2]

// Now access z by
path[i].z

Which is best for a list of points: an array of point structs or a 2 dimensional array?

2 Answers2