Optimize the mod operation

[chenquan]

Optimizing mod operation with bit operation

[chenquan]

@orcaman Maybe you can have a look?

[orcaman]

Thanks, you @chenquan, I learned a few things reading more about this type of optimization!

I would expect that the Go compiler apply these types of optimizations under the hood. Did you check if this change actually improves performance? It would be cool to include a benchmark to see if this actually moves the needle.

[orcaman]

BTW, modulo of powers of 2 can alternatively be expressed as a bitwise AND operation, but not for all numbers. Since the number of shards is an easy-to-change constant, this might cause users of this repo issues if they are not aware). Anyway thanks for this