An economical server application is crucial for both player experience and minimizing operation costs, but before optimizing anything, we need a way to measure by how much our changes improve the performance, if at all.
For this, the absolute minimum requirement is recording (or hand-editing) and replaying scenarios. In the simplest case we can just initialize a bunch of shapes with random speeds and positions, but something more realistic, like spawning bullets at a somewhat reasonable target.
From there, we can take 2 ways (ideally both):
simulate a given number of frames, measure how much time it takes
run the simulation for some time, record the remaining time in every frame, then derive various metrics like average, minimum, etc.
An economical server application is crucial for both player experience and minimizing operation costs, but before optimizing anything, we need a way to measure by how much our changes improve the performance, if at all.
For this, the absolute minimum requirement is recording (or hand-editing) and replaying scenarios. In the simplest case we can just initialize a bunch of shapes with random speeds and positions, but something more realistic, like spawning bullets at a somewhat reasonable target.
From there, we can take 2 ways (ideally both):