Sha256: 0ef3bc139b14c6f2245117ec98925a7ed4af38dba61d3fc039132fd71473eb3b
Contents?: true
Size: 1.72 KB
Versions: 35
Compression:
Stored size: 1.72 KB
Contents
#!/usr/bin/env ruby
require 'benchmark'
require 'socket'
ITERATIONS = 100000
REQUEST = " " * 800
REQUEST_SIZE = REQUEST.size
RESPONSE = " " * 5 * 1024
RESPONSE_SIZE = RESPONSE.size
PASSWORD = "x" * 128
PASSWORD_SIZE = PASSWORD.size
def start
benchmark_with_password_checking
puts ""
benchmark_without_password_checking
end
def benchmark_with_password_checking
puts "Benchmarking with password checking..."
parent, child = UNIXSocket.pair
pid = fork do
parent.close
ITERATIONS.times do
password = child.read(PASSWORD_SIZE)
if password == PASSWORD
child.read(REQUEST_SIZE)
child.write(RESPONSE)
# This flush here improves performance significantly!
child.flush
else
child.close
break
end
end
end
child.close
result = Benchmark.measure do
ITERATIONS.times do
parent.write(PASSWORD)
parent.write(REQUEST)
parent.flush
parent.read(RESPONSE_SIZE)
end
end
parent.close
Process.waitpid(pid)
puts "User/system/real time: #{result}"
printf "%.2f messages per second\n", ITERATIONS / result.real
end
def benchmark_without_password_checking
puts "Benchmarking without password checking..."
parent, child = UNIXSocket.pair
pid = fork do
parent.close
ITERATIONS.times do
child.read(REQUEST_SIZE)
child.write(RESPONSE)
# This flush here improves performance significantly!
child.flush
end
end
child.close
result = Benchmark.measure do
ITERATIONS.times do
parent.write(REQUEST)
# We do not call flush. For some reason it degrades performance.
parent.read(RESPONSE_SIZE)
end
end
parent.close
Process.waitpid(pid)
puts "Without password checking:"
puts "User/system/real time: #{result}"
printf "%.2f messages per second\n", ITERATIONS / result.real
end
start
Version data entries
35 entries across 35 versions & 5 rubygems