an "optimized hash function" ?
Last time i tried to test hash functions, i found little differences between most formulaes,
with the notable exception of byte-aligned operations, which should be avoided.
(For example, for a 4 bytes to 16 bits hash, one should avoid :
hash = input ^ input>>16;
)
That situation aside, any odd number seemed to produce mostly equivalent results;
Anyway, i'm very interested in what you could find...