Thread: Speed comparison between vector and 2*2 array

I'm refactoring someone else's code, and this code uses an underlying 1D vector to implement a 2D array. If the 2D arrays were variable in size, I'd leave it alone, but the sizes of these "arrays" are pretty much set in stone, so there's no problem changing them to straightforward 2D arrays (which I am aware are still 1D arrays in memory). Now, if I had the time, I'd do speed tests myself, but I am on a deadline, so I'm asking you experts' advice.
In a program that does several million iterations over these arrays (101 by 101 to be precise), would there be a significant speed gain if I switched from vectors to arrays?

The access speed of already allocated vectors should be identical to arrays. The vector is basically a wrapper around an array allocated with new[]. As long as the vector is not being resized all the time you should see no difference.

I haven't done any of my own speed comparisons, but from what I've read I'd agree with Brewbuck.

Chapter 76. Use vector by default. Otherwise, choose an appropriate container

Chapter 77. Use vector and string instead of arrays

Chapter 78. Use vector (and string::c_str) to exchange data with non-C++ APIs

Thanks. Good to know that I can get speed benefits while working less.

You said that the size is set in stone. That means it is known at compile time? Then you are comparing a static array with vector (as opposed to a dynamic array allocated with new[]), right?

I'd imagine the static array is actually faster. Whether the difference is noticeable or not I don't know, but if you were starting from scratch I'd say use a plain array or std::tr1::array (i.e. boost's array).