This is exactly what I am trying to do nowadays!
JohnBlackburne Most developers' main concern about memory is minimising the number of temporary objects as although the garbage collector works it is slow. So they e.g. use pools for common objects, such as the ones in Starling.util.pool or ones they create themselves.
I would like to ask based on the card game example! In most of the card games we use a `TextureAtlas' like this
where each card object is basically a single image object. Which we are considering as not optimised and consume large amount of memory(based on our current knowledge).
As a part of optimisation, we are planning to use a smaller size of texture atlas which comprise of a bitmapfont(white) and textures of ♠♥♦♣(ignoring jacks, queens & kings textures).
Now a we need to build a card object with a quad(white backround),textfiled(value), images(suits) and store each card object in a repo. If we are using multiple decks then each deck will have a separate repo to handle duplicate issue!
The question or concern is how do we measure the memory usage and corresponding performances either way!
For an example if we are using 100 decks, then there would be 5200 cards objects in the memory. Which one from the above would be then better optimised for code execution, memory performance and rendering ?