Split vs. count

bench.rb

require 'benchmark'

N = 10_000_000
S = 'string,' * N

Benchmark.bmbm do |x|
  x.report('#count') do
    (length = S.count(',')) && (length + 1).times{|i| "line #{i} of #{length}" }
  end

  x.report('#split') do
    split_s = S.split(',')
    len = split_s.length
    split_s.each_with_index{ |element, index| "line #{index} of #{len}"}
  end
end

output.txt

% ruby bench.rb
Rehearsal ------------------------------------------
#count   9.580000   0.000000   9.580000 (  9.582891)
#split  13.090000   0.520000  13.610000 ( 13.614202)
-------------------------------- total: 23.190000sec

             user     system      total        real
#count   9.050000   0.260000   9.310000 (  9.313300)
#split  11.690000   0.310000  12.000000 ( 11.999484)

With N = 100:

Rehearsal ------------------------------------------
#count   0.000000   0.000000   0.000000 (  0.000085)
#split   0.000000   0.000000   0.000000 (  0.000106)
--------------------------------- total: 0.000000sec

             user     system      total        real
#count   0.000000   0.000000   0.000000 (  0.000066)
#split   0.000000   0.000000   0.000000 (  0.000079)