https://drops.dagstuhl.de/storage/00lipics/lipics-vol301-sea...
It's the fastest sublinear search algorithm around in almost all cases. I also have a guaranteed worst-case linear version (which is still sublinear and much faster than Boyer Moore in the average case).
Sample code is available here:
https://github.com/nishihatapalmer/HashChain
If you're interested, let me know.
The ones parent picked are not amongst the best at all. I added a ticket to add your HashChain algos to be tested also.
Preliminary testing shows indeed excellent numbers for hc2-hc4. Comes close to musl
Can speed up most of the algorithms in smart by using that. It's faster than using array indexing generally. I didn't do that in the sample code for HashChain simply because it would be an unfair comparison with the others.