hypergraph_performance_tpc_test.rb

require 'test_helper'
require 'benchmark'

class HypergraphPerformanceTest < Minitest::Test
  include Hypergraph::TestHelper::Schema
  include Hypergraph::TestHelper::Benchmarking
  include Hypergraph::TestHelper::Profiling

  def setup
    suites = (ENV['TESTS'] || '').split
    suites = %w[ 5k ] if suites.length < 1

    @suites = suites.inject({}) do |h, name|
      db_name = "hypergraph_test_#{name}"
      schema = Hypergraph::Schema.new(db: connection_string(name: db_name))
      recreate_schema!(schema)
      h[name] = { schema: schema }
      h
    end

    # query context
    @dates = %w[ 2015-06-01 ].map {|s| Time.parse(s) }
    @fact_id = 1
    @scenario_id = 1
  end

  def test_performance
    with_benchmark('### Load Tables') do |name, hash|
      schema = hash[:schema]
      schema.truncate!
      schema.load!("test/data/csv/#{name}")
    end

    with_benchmark('### Instantiate Graph') do |name, hash|
      schema = hash[:schema]
      options = {
        index: {
          keys: %i[ name ]
        }
      }
      graph = nil
      profile do
        graph = hash[:graph] = Hypergraph.from_schema(schema, options)
      end

      graph.measures.derive('Total Sales') do |node, scenario_id, t|
        # Query context.
        dates    = [t]
        measures = ['Sales']

        ## Implementation A.
        ## Multithreaded with ThreadSafe::Array and inject.
        #q = ThreadSafe::Array.new
        #graph.nodes.peach(Hypergraph.threads) do |n|
	#  q << n.measures.get(scenario_id, dates, measures)[0][0]
        #end
        #q.compact.inject(:+)

        ## Implementation B.
        ## Multithreaded with mutex and +=.
        #s = 0
        #m = Mutex.new
        #graph.nodes.peach(Hypergraph.threads) do |n|
	#  x = n.measures.get(scenario_id, dates, measures)[0][0]
        #  next if x.nil?
	#  m.synchronize { s += x }
        #end
        #s

        ## Implementation C.
        ## Single-threaded.
        #s = 0
        #graph.nodes.each do |n|
	#  x = n.measures.get(scenario_id, dates, measures)[0][0]
        #  s += x unless x.nil?
        #end
        #s
      end
    end

    with_benchmark('### Query Measures (Derived)') do |name, hash|
      graph = hash[:graph]
      measures = ['Sales', 'Total Sales']
      puts "[#{measures.join(', ')}] x [#{@dates.join(', ')}]: "
      p graph[@fact_id].measures.get(@scenario_id, @dates, measures)
    end
  end

  private

  def with_benchmark(label = '', &block)
    raise ArgumentError unless block_given?
    puts label
    puts
    @suites.each do |name, hash|
      GC.start
      print name.to_s.rjust(10), " " * 10
      # Benchmark.measure reports user/system/user+system/elapsed times.
      puts Benchmark.measure(label) { yield(name, hash) }
    end
    puts
  end
end