Top 10: JavaScript, Java, Python, Ruby, Php, C++, C#, C, Go reference

Others: Crystal, Rust, Swift, Mono, Elixir, Perl, R, Julia, D, Nim, Pascal, Fortran, Cython, Pony, OCaml, Lisp, Haskell, Erlang, Escript, Dart, Clojure, Scheme, Lua, Python3, Perl, Perl6, Bash, Emoji

The code performs a recursive fibonacci to the 46th position with the result of 2,971,215,073. This is the original version where the sequence starts at 1 instead of 0. 1,1,2,3,5,8...

Fibonacci can be written many different ways. The goal of this project is to compare how each language handles the exact same code.

Here is the Ruby version:

def fib(n)
  return 1 if n <= 1
  fib(n - 1) + fib(n - 2)
end

puts fib(46)

Here is the Crystal version:

def fib(n : UInt64)
  return 1_u64 if n <= 1
  fib(n - 1) + fib(n - 2)
end

puts fib(46)

All tests are run on:

• MacBook Pro (16-inch, 2019)
• Processor: 2.6 GHz 6-Core Intel Core i7
• Memory: 32 GB 2667 MHz DDR4
• OS: macOS Mojave 10.15.5
• Docker Version: 19.03.8
• Docker Image: ubuntu:18.04

You can run the tests using Docker: docker run -it drujensen/fib

Last benchmark was ran on June 21, 2020

NOTE: These languages include compilation time which should be taken into consideration when comparing.

NOTE: Interpreted languages have a startup time cost that should be taken into consideration when comparing.

The code examples provided in the optimized folder use techniques that break the benchmark. They do not perform the same internal tasks as the other examples so are not a good for an apples to apples comparison. They all perform at sub-second response times. It demonstrates that all benchmarks will have some caveat.

Several of these examples add very little changes to the original code that are worth mentioning:

• The C++ constexpr is using a constexpr which optimizes the recursive call to a constant.
• The Lisp compiletime shows how you can perform the calculation at compile time. simply add #. before outputing the results.
• The Python lru_cache is using lru_cache directive with no code changes.
• The Ruby mem, Crystal mem and JS mem are maintaining a simple cache to avoid recalculating the results for that value.

For the fun of it, here are the benchmarks for the optimized versions

Last benchmark was ran on June 21, 2020

All compilers are installed using asdf or apt on Ubuntu 18.04 docker image: assembly (gcc) 7.4.0 bash 5.0.0 crystal 0.35.0 clojure 1.10.1 cython 0.29.15 D (lcd2) v1.20.1 dart 2.7.2 dotnet-core 3.1.201 elixir 1.10.0 elm 0.19.1 erlang 22.3.1 g++ 7.4.0 gcc 7.4.0 gfortran 9.3.0 golang 1.14.1 haskell 8.10.1 java adopt-openjdk-14+36 julia 1.4.0 K 3.6 lisp (sbcl) 2.0.3 lua 5.3.0 mono 6.8.0 nim v1.2.0 nodejs 12.16.1 ocaml 4.10.0 pascal 3.0.4 perl 5.30.2 php 7.4.4 ponyc 0.33.2 powershell v7.0.0 python 3.8.2 R 3.6.3 rakudo 2020.01 ruby 2.7.1 rust 1.42.0 scheme (guile) 2.2 swift 5.2.1 tcl 8.6.10

Fibonacci Benchmark