> Yes, although even here, small is not that very small: > > The main focus is about short keys of random lengths, with a distribution of length roughly in th

> Yes, although even here, small is not that very small: > > The main focus is about short keys of random lengths, with a distribution of length roughly in the 20-30 bytes area, featuring occasional outliers, both tiny and large. Based on the graphs, XXH3 still seems to beat everything at key sizes > 6 (presumably bytes?)