I timed each implementation in solving 6-Queens[1] on a meta-circular Scheme[2] (i.e. each Scheme interpreter ran a meta-circular Scheme interpreter to solve 6-Queens).
Results are shown in [3]; the order of speed is observed as follows:
Go ≈ Java ≈ Crystal(Compiled) ≈ SBCL > C# ≈ PyPy > TypeScript(Node.js) ≈ Crystal ≈ Dart >> PHP > Python ≈ Ruby
[1] https://github.com/nukata/little-scheme/blob/v1.2.0/examples...
[2] https://github.com/nukata/little-scheme/blob/v1.2.0/scm.scm
[3] https://github.com/nukata/little-scheme/tree/v1.2.0#performa...