benchmark comparison of Sets vs Hashes

results.txt

N = 10000000
SET INSERTION:
       user     system      total        real
  13.080000   0.370000  13.450000 ( 13.453529)
HASH INSERTION:
       user     system      total        real
  12.870000   0.330000  13.200000 ( 13.193739)
SET DELETION:
       user     system      total        real
   9.700000   0.000000   9.700000 (  9.703152)
HASH DELETION:
       user     system      total        real
   8.300000   0.000000   8.300000 (  8.306903)
[Finished in 45.3s]

set_vs_hash.rb

# Sets are like Hashes without significant values
# One would hope that deletion by key would be constant-time, like in hashes.
# This benchmark investigates that.

require 'benchmark'
require 'set'

NUM = 10_000_000

def bench(&blk)
  Benchmark.bm { |bm| bm.report &blk }
end

def set_insertion
  NUM.times.reduce Set.new, &:<<
end

def hash_insertion
  NUM.times.each_with_object(Hash.new) { |n, hash| hash[n] = true }
end

def set_deletion(set)
  set.tap { NUM.times.each { |n| set.delete n } }
end

def hash_deletion(hash)
  hash.tap { NUM.times.each { |n| hash.delete n } }
end

puts "\nN = #{NUM}\n"
puts "SET INSERTION:\n"
bench { @set = set_insertion } 
puts "HASH INSERTION:\n"
bench { @hash = hash_insertion }
puts "SET DELETION:\n"
bench { set_deletion(@set) }
puts "HASH DELETION:\n"
bench { set_deletion(@hash) }