Well, based on this sentence:
> In this implementation, to maximise performance, we consider each individual byte of the file separately, and look it up in π.
I'd say best-case scenario, you're looking at 1:1 offset storage size vs. stored data size :)