Isn't this simply bound by the speed of the disk? Surely none of the suggested optimizations (SIMD, multi-threading) are relevant.
It would also depend of how many different stations there are and what they are for the hash lookup, but seriously I doubt this will be anything measurable compared to I/O.
It would also depend of how many different stations there are and what they are for the hash lookup, but seriously I doubt this will be anything measurable compared to I/O.