Package

purescript-benchotron

Repository
hdgarrood/purescript-benchotron
License
MIT
Uploaded by
hdgarrood

Build Status

Straightforward benchmarking via Benchmark.js. I am sorry about the name (purescript-benchmark was taken).

usage

Suppose you want to find out which is faster out of foldr (+) 0 and runAdditive <<< foldMap Additive. Let's also do the same for (*) for good measure. Start by creating some Benchmark values:

module Main where

import Prelude
import Control.Monad.Eff (Eff)
import Data.Array ((..))
import Data.Foldable (foldMap, foldr)
import Data.Monoid.Additive (Additive(..))
import Data.Monoid.Multiplicative (Multiplicative(..))
import Data.Newtype (ala)
import Test.QuickCheck.Arbitrary (arbitrary)
import Test.QuickCheck.Gen (vectorOf)
import Benchotron.Core (Benchmark, BenchEffects, benchFn, mkBenchmark)
import Benchotron.UI.Console (runSuite)

benchSum :: Benchmark
benchSum = mkBenchmark
  { slug: "sum"
  , title: "Finding the sum of an array"
  , sizes: (1..5) <#> (_ * 1000)
  , sizeInterpretation: "Number of elements in the array"
  , inputsPerSize: 1
  , gen: \n -> vectorOf n arbitrary
  , functions: [ benchFn "foldr" (foldr (+) 0)
               , benchFn "foldMap" (ala Additive foldMap)
               ]
  }

benchProduct :: Benchmark
benchProduct = mkBenchmark
  { slug: "product"
  , title: "Finding the product of an array"
  , sizes: (1..5) <#> (_ * 1000)
  , sizeInterpretation: "Number of elements in the array"
  , inputsPerSize: 1
  , gen: \n -> vectorOf n arbitrary
  , functions: [ benchFn "foldr" (foldr (*) 1)
               , benchFn "foldMap" (ala Multiplicative foldMap)
               ]
  }
main :: forall eff. Eff (BenchEffects eff) Unit
main = runSuite [benchSum, benchProduct]

Now, run them with runSuite; this will save the results data for each benchmark to tmp/sum.json and tmp/product.json respectively.

main = runSuite [benchSum, benchProduct]

You can now generate SVG graphs of these results by visiting http://harry.garrood.me/purescript-benchotron-svg-renderer.

Further information, such as the meaning of each of the fields of a Benchmark value, is available in the documentation on Pursuit.