git.iko.soy/text
on master help sign in
Initial cut at cross-language benchmarks
9eab1f7Bryan O'Sullivan
benchmarks/haskell/Multilang.hsnew file+32
compare with master
unifiedsplit
@@ -0,0 +1,32 @@
1+{-# LANGUAGE BangPatterns, OverloadedStrings, RankNTypes #-}
2+
3+module Main (
4+ main
5+ ) where
6+
7+import Control.Monad (forM_)
8+import qualified Data.ByteString as B
9+import qualified Data.Text as Text
10+import Data.Text.Encoding (decodeUtf8)
11+import Data.Text (Text)
12+import System.IO (hFlush, stdout)
13+import Timer (timer)
14+
15+type BM = Text -> ()
16+
17+bm :: forall a. (Text -> a) -> BM
18+bm f t = f t `seq` ()
19+
20+benchmarks :: [(String, Text.Text -> ())]
21+benchmarks = [
22+ ("find_first", bm $ Text.isInfixOf "en:Benin")
23+ , ("find_index", bm $ Text.findIndex (=='c'))
24+ ]
25+
26+main :: IO ()
27+main = do
28+ !contents <- decodeUtf8 `fmap` B.readFile "../tests/text-test-data/yiwiki.xml"
29+ forM_ benchmarks $ \(name, bmark) -> do
30+ putStr $ name ++ " "
31+ hFlush stdout
32+ putStrLn =<< (timer 100 contents bmark)
benchmarks/haskell/Timer.hsnew file+30
@@ -0,0 +1,30 @@
1+{-# LANGUAGE BangPatterns #-}
2+
3+module Timer (timer) where
4+
5+import Control.Exception (evaluate)
6+import Data.Time.Clock.POSIX (getPOSIXTime)
7+import GHC.Float (FFFormat(..), formatRealFloat)
8+
9+ickyRound :: Int -> Double -> String
10+ickyRound k = formatRealFloat FFFixed (Just k)
11+
12+timer :: Int -> a -> (a -> b) -> IO String
13+timer count a0 f = do
14+ let loop !k !fastest
15+ | k <= 0 = return fastest
16+ | otherwise = do
17+ start <- getPOSIXTime
18+ let inner a i
19+ | i <= 0 = return ()
20+ | otherwise = evaluate (f a) >> inner a (i-1)
21+ inner a0 count
22+ end <- getPOSIXTime
23+ let elapsed = end - start
24+ loop (k-1) (min fastest (elapsed / fromIntegral count))
25+ t <- loop (3::Int) 1e300
26+ let log10 x = log x / log 10
27+ ft = realToFrac t
28+ prec = round (log10 (fromIntegral count) - log10 ft)
29+ return $! ickyRound prec ft
30+{-# NOINLINE timer #-}
benchmarks/python/multilang.pynew file+50
@@ -0,0 +1,50 @@
1+#!/usr/bin/env python
2+
3+import math
4+import sys
5+import time
6+
7+def find_first():
8+ cf = contents.find
9+ return timer(lambda: cf("en:Benin"))
10+
11+def timer(f, count=100):
12+ a = 1e300
13+ def g():
14+ return
15+ for i in xrange(3):
16+ start = time.time()
17+ for j in xrange(count):
18+ g()
19+ a = min(a, (time.time() - start) / count)
20+
21+ b = 1e300
22+ for i in xrange(3):
23+ start = time.time()
24+ for j in xrange(count):
25+ f()
26+ b = min(b, (time.time() - start) / count)
27+
28+ return round(b - a, int(round(math.log(count, 10) - math.log(b - a, 10))))
29+
30+contents = open('../../tests/text-test-data/yiwiki.xml', 'r').read()
31+contents = contents.decode('utf-8')
32+
33+benchmarks = (
34+ find_first,
35+ )
36+
37+to_run = sys.argv[1:]
38+bms = []
39+if to_run:
40+ for r in to_run:
41+ for b in benchmarks:
42+ if b.__name__.startswith(r):
43+ bms.append(b)
44+else:
45+ bms = benchmarks
46+
47+for b in bms:
48+ sys.stdout.write(b.__name__ + ' ')
49+ sys.stdout.flush()
50+ print b()
benchmarks/text-benchmarks.cabal+9
@@ -35,3 +35,12 @@ executable text-benchmarks
35 ghc-prim,
35 ghc-prim,
36 stringsearch,
36 stringsearch,
37 utf8-string
37 utf8-string
38+
39+executable text-multilang
40+ hs-source-dirs: haskell
41+ main-is: Multilang.hs
42+ ghc-options: -Wall -O2
43+ build-depends: base == 4.*,
44+ bytestring,
45+ text,
46+ time