This is the sort of question I personally find really useful; optimizing common algorithms for uncommon data sets can be a very tedious... And often just doesn't get done because unless someone's already done the leg-work it's too much of an investment. So I love seeing them on SO.
That said, you have a small problem with your question as it stands:
The accepted answer here Efficiently compute Intersection of two Sets in Java? looked extremely helpful but (likely because my sets are small (?)) I haven't gotten much improvement by using the approach suggested in the accepted answer.
Give us some more information on your sets. How small? How many of them? If someone's looking to optimize for a situation wholly different from what's been considered in the previous question, it's important that they have a good understanding of the nature of that situation! Ideally, folks would be able to test their solutions against data of a certain form (if not necessarily specific data).
Do that, and I think it'd be a perfectly good question.