Some languages have hash tables (and hence sets, as they tend to be implemented with them) that use open addressing, in a way that doesn't end up being bad cach

Some languages have hash tables (and hence sets, as they tend to be implemented with them) that use open addressing, in a way that doesn't end up being bad cachewise unless there are unreasonably many collisions for the same hash code. It is also not uncommon for mature implementations to optimize the cases with few elements in the hash to use linear lookup. In some cases that optimization is also used while storing data in small hash tables. Pointer chasing hash tables was good when cache was nonexistant or small (ie the 90s), but nowadays, open addressing is just superior.