ruby_sets_benchmark.rb

require 'benchmark'
require 'set'
require 'fds'

Benchmark.bm(30) do |bm|
  bm.report('Using FDS::UnorderedSet') do
    set = FDS::UnorderedSet.new
    10_000.times do
      set << rand
      set.find_index(42)
    end
  end

  bm.report('Using default Ruby Set') do
    set = Set.new
    10_000.times do
      set << rand
      set.find_index(42)
    end
  end
end

#                                      user     system      total        real
# Using FDS::UnorderedSet          0.010000   0.000000   0.010000 (  0.006958)
# Using default Ruby Set           4.030000   0.010000   4.040000 (  4.059034)

Minor nit: but the execution time of puts "Test 0: #{end_test_0 - start}" is getting added to the second set's benchmark. Might be better to have have to separate start variables just to keep things apples to apples.

@rringler, true. I just updated the gist using recommended gem Benchmark. The result should be closer to reality. 😄
One should note that the slow operation for Ruby Set here is Enumerable#find_index. Insertion is about 15% slower on FDS implementation (because of hash values initialization).