Technically, they're wrapping a native library, so it's possible to create a benchmark where there's no performance gap whatsoever. The bottleneck is at the int

Technically, they're wrapping a native library, so it's possible to create a benchmark where there's no performance gap whatsoever. The bottleneck is at the intersection of the runtime enviroment with the library, which makes comparison difficult. FWIW it's a good piece of code, and i like it.