useless microbenchmark

bench.md

For ruby:

require 'benchmark'
puts Benchmark.measure {(1..100_000).reduce(&:*); nil}.real

I get 7.48 to 7.68 seconds as the runtime on ruby 2.6.3 and 7.71 to 8.08 for ruby 3.0.0.

For elixir:

defmodule Factorial do
  def run do
    Enum.reduce(1..100_000, 1, & &1 * &2)
    :ok
  end
end

micros = :timer.tc(Factorial, :run, []) |> elem(0)
IO.puts "#{micros / 1_000_000} seconds"

I get 4.68 to 4.71 seconds for Erlang/OTP 24 RC2 with Elixir 1.11.4.

for ruby I get 3.9 sec
for elixir I get 4.891474 sec
in my apple m1 air (16g)

@coolmenu I'm curious what build versions you're using and if you're using rosetta 2. I don't have an M1 Mac to try this with but I'd expect the erlang JIT to work quite well with it as it generates code during startup so translation caches can be effective. You might see it drop well below ruby's runtime. If you're using homebrew + rosetta you can install the JIT version with brew install --HEAD erlang.

For hardware info, I'm running this on an 8-core i9 with macOS 11.2.3.

Ran both in the interactive shells. Here is what I got:

1.052038 seconds for Elixir (Erlang/OTP 26 [erts-14.0.2])
4.762765 seconds for Ruby (irb 1.6.2)

Machine is intel-based Macbook Pro 32GB 8 core i9.