With vector for trivial things, push_back is almost always slower than resizing and then trimming down if needed. The memset to zero it out is minuscule to the

With vector for trivial things, push_back is almost always slower than resizing and then trimming down if needed. The memset to zero it out is minuscule to the cost of push_back. Guess if you don't know how many and it will probably be faster. if you run out, do another block