Micro bench for SO http://stackoverflow.com/questions/15771293/selecting-values-from-a-large-hash-based-on-a-large-array-in-ruby-is-slow

hash_array_mask.bm.rb

#!/usr/bin/env ruby

require 'set'
require 'benchmark'

N = 5

HSZ = 150_000
ASZ =  25_000

HH = Hash[ (0...HSZ).map{|i| [i, 'something'] } ].freeze
A = (0...ASZ).map{ Random.rand(HSZ) }.freeze
S = A.to_set.freeze

# to account for the HH.dup cost reseting hash on each iteration.
def noop(h,a,s)
end

def keys_minus_arr_delete(h,a,s)
  (h.keys - a).each{|k| h.delete(k)}
end

def select_set_include(h,a,s)
  h.select{|k,_| s.include?(k) }
end

def select_set_include_fly(h,a,s)
  s = a.to_set
  h.select{|k,_| s.include?(k) }
end

def select_arr_include(h,a,s)
  h.select{|k,_| a.include?(k) }
end

def sawa2(h,a,s)
  keys_to_be_removed = {}
  h.each{|k, _| keys_to_be_removed[k] = true}
  a.each{|k| keys_to_be_removed[k] = false}
  keys_to_be_removed.each{|k, v| h.delete(k) if v}
end

methods = [:noop, :keys_minus_arr_delete, :select_set_include, :select_set_include_fly, :sawa2
# :select_arr_include,   # takes too long
]

Benchmark.bm(25) do |x|
  methods.each do |m|
    x.report(m.to_s) { N.times{ send(m, HH.dup, A, S) } }
  end
end