e1==e2
is too cheap to make it considerably faster with parallelization. But actually, the Bag
method contains already too much knowledge and care for memory. We know/hope that at least some elements will be removed from the list and therefore, we rather use a dynamically growing list with Bag
.