It's much faster than std::map and in many cases even faster than std::unordered_map. For example inserts of 10M random keys (16 bytes each) processed within 1.3 sec but std::map spent on the same test about 14 secs.

In contrast to hashtables it has more wide functionality:

- Prefix compression

- Ordered keys. Possibility create custom ordering

- Scan ranges of keys, sub ranges of keys

- Very fast serialization\desreialization on disc.

- Fair delete keys with smoothly releasing memory

Codebase with benchmarks available here

https://github.com/Bazist/HArray