1brc_processes.rb

# frozen_string_literal: true

require 'json'

SEMI = String.new(';', encoding: Encoding::BINARY)
NEWLINE = String.new("\n", encoding: Encoding::BINARY)

FILENAME = 'measurements.txt'
NUMPARTS = 16

endpos = File.size(FILENAME)
partsize = (endpos / NUMPARTS).ceil

pipe_out, pipe_in = IO.pipe

topfile = File.open(FILENAME, encoding: Encoding::BINARY)

(NUMPARTS - 1).downto(0).map do |part_index| # rubocop:disable Metrics/BlockLength
  partstartpos = 0
  if part_index.positive?
    topfile.seek(endpos - partsize)
    topfile.readline(NEWLINE)
    partstartpos = topfile.pos
  end

  partendpos = endpos
  endpos = topfile.pos
  fork do
    pipe_out.close
    filepart = File.open(FILENAME, encoding: Encoding::BINARY)
    filepart.seek(partstartpos)
    stats = Hash.new { |h, k| h[k] = [10_000, -10_000, 0, 0] }
    loop do
      city = filepart.readline(SEMI)
      temp = filepart.readline(NEWLINE).to_f
      existing = stats[city]
      min, max, sum, count = existing
      if temp < min
        existing[0] = temp
      elsif temp > max
        existing[1] = temp
      end
      existing[2] = sum + temp
      existing[3] = count + 1

      # endpos should be exact, but still
      break if filepart.pos >= partendpos
    end
    pipe_in.puts JSON.generate(stats)
  rescue EOFError
    # all good
  ensure
    filepart.close
  end
end
pipe_in.close

stats = {}.merge(*pipe_out.readlines.map { |l| JSON.parse(l) }) do |_, s1, s2|
  [
    s1[0] < s2[0] ? s1[0] : s2[0],
    s1[1] > s2[1] ? s1[1] : s2[1],
    s1[2] + s2[2],
    s1[3] + s2[3]
  ]
end

puts JSON.pretty_generate(stats.sort.map do |(city, cstats)|
  "#{city[0..-2]}=#{cstats[0]}/#{format('%0.1f', (cstats[2] / cstats[3]))}/#{cstats[1]}"
end)

1brc_ractors.rb

FILENAME = 'measurements.txt'
NUMPARTS = 8

endpos = File.size(FILENAME)
partsize = (endpos / NUMPARTS).ceil

threads = (NUMPARTS - 1).downto(0).map do |part_index| # rubocop:disable Metrics/BlockLength
  filepart = File.open(FILENAME, encoding: Encoding::BINARY)
  if part_index.positive?
    filepart.seek(endpos - partsize)
    filepart.readline("\n")
  end
  partendpos = endpos
  endpos = filepart.pos
  Ractor.new(filepart, partendpos) do |filepart, partendpos|
    semi = String.new(';', encoding: Encoding::BINARY)
    newline = String.new("\n", encoding: Encoding::BINARY)

    stats = Hash.new { |h, k| h[k] = [10_000, -10_000, 0, 0] }
    loop do
      city = filepart.readline(semi)
      temp = filepart.readline(newline).to_f
      existing = stats[city]
      min, max, sum, count = existing
      if temp < min
        existing[0] = temp
      elsif temp > max
        existing[1] = temp
      end
      existing[2] = sum + temp
      existing[3] = count + 1

      # endpos should be exact, but still
      break if filepart.pos >= partendpos
    end
    stats
  rescue EOFError
    # all good
  ensure
    filepart.close
  end
end

stats = {}.merge(*threads.map(&:take)) do |_, s1, s2|
  [
    s1[0] < s2[0] ? s1[0] : s2[0],
    s1[1] > s2[1] ? s1[1] : s2[1],
    s1[2] + s2[2],
    s1[3] + s2[3]
  ]
end

require 'json'

puts JSON.pretty_generate(stats.sort.map do |(city, cstats)|
  "#{city[0..-2]}=#{cstats[0]}/#{format('%0.1f', (cstats[2] / cstats[3]))}/#{cstats[1]}"
end)

1brc_threads.rb

SEMI = String.new(';', encoding: Encoding::BINARY)
NEWLINE = String.new("\n", encoding: Encoding::BINARY)

FILENAME = 'measurements.txt'
NUMPARTS = 8

endpos = File.size(FILENAME)
partsize = (endpos / NUMPARTS).ceil

threads = (NUMPARTS - 1).downto(0).map do |part_index| # rubocop:disable Metrics/BlockLength
  filepart = File.open(FILENAME, encoding: Encoding::BINARY)
  if part_index.positive?
    filepart.seek(endpos - partsize)
    filepart.readline(NEWLINE)
  end
  partendpos = endpos
  endpos = filepart.pos
  Thread.new do
    stats = Hash.new { |h, k| h[k] = [10_000, -10_000, 0, 0] }
    loop do
      city = filepart.readline(SEMI)
      temp = filepart.readline(NEWLINE).to_f
      existing = stats[city]
      min, max, sum, count = existing
      if temp < min
        existing[0] = temp
      elsif temp > max
        existing[1] = temp
      end
      existing[2] = sum + temp
      existing[3] = count + 1

      # endpos should be exact, but still
      break if filepart.pos >= partendpos
    end
    stats
  rescue EOFError
    # all good
  ensure
    filepart.close
  end
end

stats = {}.merge(*threads.map(&:value)) do |_city, s1, s2|
  [
    s1[0] < s2[0] ? s1[0] : s2[0],
    s1[1] > s2[1] ? s1[1] : s2[1],
    s1[2] + s2[2],
    s1[3] + s2[3]
  ]
end

require 'json'

puts JSON.pretty_generate(stats.sort.map do |(city, cstats)|
  "#{city[0..-2]}=#{cstats[0]}/#{format('%0.1f', (cstats[2] / cstats[3]))}/#{cstats[1]}"
end)