I’m trying to come up with an accurate measure of how much memory a particular datastructure will consume. Big part of that is ~ 1Million Uint32Array and 1M BigUint64Array both of max 200 elements.
Is my thinking correct that typed arrays that are initialised to a particular size already consume memory even without elements having been inserted, and that inserts don’t change the amount of memory allocated?
If so, I can quickly get a sense of memory needed.
>Solution :
Yes. Constructing a new typed array initialises it with a buffer of the respective size, which allocates the memory.
In theory, an engine could defer the allocation until the buffer is actually used to store values, but I haven’t seen such an optimisation anywhere. It probably happens on the OS level anyway when paging the virtual memory.