fstd with that many tees in close proximity Iām not quite sure any naive rework of physics like chunking would help. Maybe a quadtree could have good enough performance characteristics, though those are not cheap to update so maybe not?
12:40
Would be an interesting challange to get 128 tees on a server colliding like that without turning the performance to shit
12:42
Space hashing with very small chunks could work, though with very small chunks you get a massive hashmap, might not fit in a cache line