> ...but let's please try to NOT include hash functions that do
> table lookups (especially when the table is too large or used
> too infrequently to fit nicely in the L1) .. it'd be a real
> shame to lose the benefit of a better hash function by polluting
> your data cache.

Yes - the benchmark will run quickly because it will get all the table
into the L1 cache.  In real life the table entries you want will never
be present - so every table lookup will be a cache miss (not only slow
but also displacing something else as well!)

While you are at it, remember NOT to unroll your loops, don't want
all that code displaced from the I-cache either......

    David