I am well aware of all the problems involved in comparing floats. This is exactly the reason for this question.
I\'m looking to create a fast hash table for values that
Do you need it to be a fast hash table or would a tree structure do?
It seems to me that it would be easier to look up matching floats in a search tree of some sort. A B-Tree minimizes the number of cache misses, assuming that you choose the right node size. That ought to make it pretty fast in practice.