Comment by kloeq

Yes the aim is to improve varint and I'm not the only one. This algorithm is the fastest I've seen yet. Also FLIT64 works well with common CPU operations.