Raw pointer, smart pointer or std::vector for "low-level" container data in C++

Question

Let's say I am making my own Matrix class/container which itself needs to manage some sort of array of doubles and a row/column dimension.

At the risk of sounding hand-wavy, what is the considered "best practice" for how to store this data if speed is of importance? The options I can see are:

• Raw pointer to single dynamic C array
• Unique pointer to a C array, which is similar/leads us to...
• std::vector of doubles

What are the speed implications of these different options? Obviously it does depend on circumstance, but in a general case? Also, the size of an std::vector on MSVC and GCC for me is 24 bytes, indicating 3 pointers to the begin iterator, end iterator and the end of the memory allocation. Since I need to store the size myself to be aware of the Matrix dimensions, the end iterator is somewhat useless to me, expect for use with algorithms.

What are the thoughts on best practices of this? Is using a raw pointer acceptable since the container is somewhat "low-level"?

Thanks!

question from:https://stackoverflow.com/questions/65642751/raw-pointer-smart-pointer-or-stdvector-for-low-level-container-data-in-c

Answer 1

I would use std::vector because it solves memory allocation, deallocation, indexing, copying, etc.. Unless you will be using "millions" of matrices at the same time, the extra member (capacity) is probably not relevant.

In any case, optimizing the library for speed is the last thing you want to do -- after you can test the actual speed of your initial implementation. Then you can decide if it is worth spending time to effectively duplicate std::vector functionality with your own implementation.