7dac6e9 · text · git.iko.soy

git.iko.soy/text

Backport integer builder benchmarks

commit7dac6e961a6b45cc42b3412974bfe7de007ab591 authorBryan O'Sullivan <bos@serpentine.com> dateTue 19 Mar 2013, 22:28 (14 years ago) parent3700f70

6 files changed+61−19

compare with master

unifiedsplit

Data/Text/Encoding/Error.hs+5 benchmarks/haskell/Benchmarks.hs+2−1 benchmarks/haskell/Benchmarks/Builder.hs+31−8 benchmarks/haskell/Benchmarks/DecodeUtf8.hs+1−1 benchmarks/haskell/Benchmarks/Pure.hs+14−8 benchmarks/text-benchmarks.cabal+8−1

Data/Text/Encoding/Error.hs+5

@@ -35,6 +35,7 @@ module Data.Text.Encoding.Error

3535 , replace

3636 ) where

3737

38+import Control.DeepSeq (NFData (..))

3839#if __GLASGOW_HASKELL__ >= 610

3940import Control.Exception (Exception, throw)

4041#else

@@ -94,6 +95,10 @@ instance Show UnicodeException where

9495

9596instance Exception UnicodeException

9697

98+instance NFData UnicodeException where

99+ rnf (DecodeError desc w) = rnf desc `seq` rnf w `seq` ()

100+ rnf (EncodeError desc c) = rnf desc `seq` rnf c `seq` ()

101+

97102-- | Throw a 'UnicodeException' if decoding fails.

98103strictDecode :: OnDecodeError

99104strictDecode desc c = throw (DecodeError desc c)

benchmarks/haskell/Benchmarks.hs+2−1

@@ -49,7 +49,8 @@ benchmarks = do

4949 , Equality.benchmark (tf "japanese.txt")

5050 , FileRead.benchmark (tf "russian.txt")

5151 , FoldLines.benchmark (tf "russian.txt")

52− , Pure.benchmark (tf "japanese.txt")

52+ , Pure.benchmark "tiny "(tf "tiny.txt")

53+ , Pure.benchmark "japanese" (tf "japanese.txt")

5354 , ReadNumbers.benchmark (tf "numbers.txt")

5455 , Replace.benchmark (tf "russian.txt") "принимая" "своем"

5556 , Search.benchmark (tf "russian.txt") "принимая"

benchmarks/haskell/Benchmarks/Builder.hs+31−8

@@ -20,18 +20,41 @@ import qualified Data.ByteString.Lazy as LB

2020import qualified Data.Text as T

2121import qualified Data.Text.Lazy as LT

2222import qualified Data.Text.Lazy.Builder as LTB

23+import qualified Data.Text.Lazy.Builder.Int as Int

24+import Data.Int (Int64)

2325

2426benchmark :: IO Benchmark

2527benchmark = return $ bgroup "Builder"

26− [ bench "LazyText" $ nf

27− (LT.length . LTB.toLazyText . mconcat . map LTB.fromText) texts

28− , bench "Binary" $ nf

29− (LB.length . B.toLazyByteString . mconcat . map B.fromByteString)

30− byteStrings

31− , bench "Blaze" $ nf

32− (LB.length . Blaze.toLazyByteString . mconcat . map Blaze.fromString)

33− strings

28+ [ bgroup "Comparison"

29+ [ bench "LazyText" $ nf

30+ (LT.length . LTB.toLazyText . mconcat . map LTB.fromText) texts

31+ , bench "Binary" $ nf

32+ (LB.length . B.toLazyByteString . mconcat . map B.fromByteString)

33+ byteStrings

34+ , bench "Blaze" $ nf

35+ (LB.length . Blaze.toLazyByteString . mconcat . map Blaze.fromString)

36+ strings

37+ ]

38+ , bgroup "Int"

39+ [ bgroup "Decimal"

40+ [ bgroup "Positive" .

41+ flip map numbers $ \n ->

42+ (bench (show (length (show n))) $ nf (LTB.toLazyText . Int.decimal) n)

43+ , bgroup "Negative" .

44+ flip map numbers $ \m ->

45+ let n = negate m in

46+ (bench (show (length (show n))) $ nf (LTB.toLazyText . Int.decimal) n)

47+ ]

48+ ]

3449 ]

50+ where

51+ numbers :: [Int64]

52+ numbers = [

53+ 6, 14, 500, 9688, 10654, 620735, 5608880, 37010612,

54+ 731223504, 5061580596, 24596952933, 711732309084, 2845910093839,

55+ 54601756118340, 735159434806159, 3619097625502435, 95777227510267124,

56+ 414944309510675693, 8986407456998704019

57+ ]

3558

3659texts :: [T.Text]

3760texts = take 200000 $ cycle ["foo", "λx", "由の"]

benchmarks/haskell/Benchmarks/DecodeUtf8.hs+1−1

@@ -18,7 +18,7 @@ module Benchmarks.DecodeUtf8

1818 ( benchmark

1919 ) where

2020

21−import Foreign.C.Types (CInt, CSize)

21+import Foreign.C.Types

2222import Data.ByteString.Internal (ByteString(..))

2323import Foreign.Ptr (Ptr, plusPtr)

2424import Foreign.ForeignPtr (withForeignPtr)

benchmarks/haskell/Benchmarks/Pure.hs+14−8

@@ -4,7 +4,7 @@

44--

55-- * Most pure functions defined the string types

66--

7−{-# LANGUAGE BangPatterns, GADTs, MagicHash #-}

7+{-# LANGUAGE BangPatterns, CPP, GADTs, MagicHash #-}

88{-# OPTIONS_GHC -fno-warn-orphans #-}

99module Benchmarks.Pure

1010 ( benchmark

@@ -27,8 +27,8 @@ import qualified Data.Text.Lazy as TL

2727import qualified Data.Text.Lazy.Builder as TB

2828import qualified Data.Text.Lazy.Encoding as TL

2929

30−benchmark :: FilePath -> IO Benchmark

31−benchmark fp = do

30+benchmark :: String -> FilePath -> IO Benchmark

31+benchmark kind fp = do

3232 -- Evaluate stuff before actually running the benchmark, we don't want to

3333 -- count it here.

3434

@@ -100,6 +100,10 @@ benchmark fp = do

100100 , benchBSL $ nf BL.unpack bla

101101 , benchS $ nf UTF8.toString bsa

102102 ]

103+ , bgroup "decode'"

104+ [ benchT $ nf T.decodeUtf8' bsa

105+ , benchTL $ nf TL.decodeUtf8' bla

106+ ]

103107 , bgroup "drop"

104108 [ benchT $ nf (T.drop (ta_len `div` 3)) ta

105109 , benchTL $ nf (TL.drop (tla_len `div` 3)) tla

@@ -405,11 +409,11 @@ benchmark fp = do

405409 ]

406410 ]

407411 where

408− benchS = bench "String"

409− benchT = bench "Text"

410− benchTL = bench "LazyText"

411− benchBS = bench "ByteString"

412− benchBSL = bench "LazyByteString"

412+ benchS = bench ("String+" ++ kind)

413+ benchT = bench ("Text+" ++ kind)

414+ benchTL = bench ("LazyText+" ++ kind)

415+ benchBS = bench ("ByteString+" ++ kind)

416+ benchBSL = bench ("LazyByteString+" ++ kind)

413417

414418 c = 'й'

415419 p0 = (== c)

@@ -425,11 +429,13 @@ benchmark fp = do

425429 replicat n = concat . L.replicate n

426430 short = T.pack "short"

427431

432+#if !MIN_VERSION_bytestring(0,10,0)

428433instance NFData BS.ByteString

429434

430435instance NFData BL.ByteString where

431436 rnf BL.Empty = ()

432437 rnf (BL.Chunk _ ts) = rnf ts

438+#endif

433439

434440data B where

435441 B :: NFData a => a -> B

benchmarks/text-benchmarks.cabal+8−1

@@ -15,13 +15,19 @@ build-type: Simple

1515

1616cabal-version: >=1.2

1717

18+flag llvm

19+ description: use LLVM

20+ default: False

21+

1822executable text-benchmarks

1923 hs-source-dirs: haskell ..

2024 c-sources: ../cbits/cbits.c

2125 cbits/time_iconv.c

2226 main-is: Benchmarks.hs

2327 ghc-options: -Wall -O2

24− cpp-options: -DHAVE_DEEPSEQ

28+ if flag(llvm)

29+ ghc-options: -fllvm

30+ cpp-options: -DHAVE_DEEPSEQ -DINTEGER_GMP

2531 build-depends: base == 4.*,

2632 binary,

2733 blaze-builder,

@@ -33,6 +39,7 @@ executable text-benchmarks

3339 directory,

3440 filepath,

3541 ghc-prim,

42+ integer-gmp,

3643 stringsearch,

3744 utf8-string

3845