I've some problem with site rendering on Safari on my MacBook Pro (apparently
the flags to filter results with / without parallelization do not work), but
not with Android smartphone. It seems that a lots of Rust results are with
parallelization, while Nim has just one for spectral-norm (much slower than
Rust ones). If you look on results without parallelization, Rust results are
still better most of times, but in the same order of magnitude (yeah sometimes
2x, that is much faster... but it's still the same order of magnitude) and in
many cases fairly close. In addition, in many benchmarks we find multiple
versions of same test with Rust (even with the same compiler) vs. just one for
Nim, so I think that there is a little bias, at least in terms of programmer
proficiency with the two languages. While performance is a big selling point of
any system language, it's not this kind of biased - muscular benchmarks that
should orient people's decision between Rust and Nim. In my opinion the
original (quite old - 2015) blog post captures better the philosophy and the
strength / weakness of the two languages.
