Writing a blog post on Data Oriented Design. Any skeptics want to share their worries/criticisms? Looking to address some of the most common ones.
Näytä tämä ketju
-
Affinities: across two frames, a task may run on two separate threads each frame. Your cache is long gone by then. There are a lot of cases and it’s much more than 1 tweet. The more interesting bit is affinities between tasks inside vs outside the graph.
Basically, if you have 8 cores and you spawn 32 async load/process tasks, the scheduler will reserve a configurable # of cores for the graph, and allow others to assist on loading. Makes sure the graph never gets choked out (gotta maintain fps).