Ruby Digest::MD5 update vs hexdigest comparison

test.rb

#!/usr/bin/env ruby -w
# frozen_string_literal: true

require 'bundler/inline'

gemfile(true) do
  source 'https://rubygems.org'
  gem 'benchmark-memory'
end

require 'benchmark'
require 'benchmark/memory'
require 'securerandom'
require 'digest/md5'

ITERATIONS = 1_000
STRING_COUNT = 10_000
STRING_PARTS = Array.new(STRING_COUNT).map { SecureRandom.alphanumeric(100) }
SINGLE_STRING = STRING_PARTS.join("\n")

Benchmark.bmbm do |b|
  b.report('update') do
    ITERATIONS.times do
      digest = Digest::MD5.new
      STRING_PARTS.each { |p| digest.update(p) }
      digest.hexdigest
    end
  end

  b.report('single large string') do
    ITERATIONS.times do
      Digest::MD5.hexdigest(SINGLE_STRING)
    end
  end
end

Benchmark.memory do |b|
  b.report('update') do
    ITERATIONS.times do
      digest = Digest::MD5.new
      STRING_PARTS.each { |p| digest.update(p) }
      digest.hexdigest
    end
  end

  b.report('single large string') do
    ITERATIONS.times do
      Digest::MD5.hexdigest(SINGLE_STRING)
    end
  end

  b.compare!
end

test.results

Rehearsal -------------------------------------------------------
update                2.874825   0.001647   2.876472 (  2.877840)
single large string   1.356260   0.000497   1.356757 (  1.357062)
---------------------------------------------- total: 4.233229sec

                          user     system      total        real
update                2.806647   0.001140   2.807787 (  2.808370)
single large string   1.280238   0.000906   1.281144 (  1.282102)
Calculating -------------------------------------
              update   193.000k memsize (     0.000  retained)
                         4.000k objects (     0.000  retained)
                         2.000  strings (     0.000  retained)
 single large string   153.000k memsize (     0.000  retained)
                         3.000k objects (     0.000  retained)
                         2.000  strings (     0.000  retained)

Comparison:
 single large string:     153000 allocated
              update:     193000 allocated - 1.26x more