Last time we talked about how we can approximate objects with complex shapes using simpler ones for our game’s physics simulation.
Further, we saw how we can use an often already existing feature: a skeleton for animating sprites – or vertices of a 3d-mesh – to make our collision shapes change position, and even size, as our object deforms.
Today I want to look at how knowing about the behaviour of CPUs – and especially their caches and memory – we can use simple optimisations to implement collisions with such objects very efficiently.