On Sun, Dec 18, 2011 at 2:57 PM, Alexey Khudyakov wrote:

Hello!

I've found a puzzling performance problem with code which uses vector library and relies heavily on GHC to perform inlining and specialization. In some cases compiler refuses to specialize function and just copies there generic version which is slow.

Here is smallest test case I've manages to make:

This is a guess, but based on what I've read the GHC inliner only fires when the function is fully saturated as declared - so if you declare a function with one argument to the left of the '=' symbol, the inliner only then inlines when it is applied to one value. This means that the un-inlined function is passed to criterion in the first case, but not the second. Does adding a SPECIALIZE pragma help? Antoine

file 'test.hs'

...

import Criterion.Main import qualified Data.Vector.Unboxed as U import Boundary

sample :: U.Vector Double sample = U.replicate 10000 0

main = defaultMain   [ bench "eta" $ nf variance sample   , bench "lambda" $ nf (\x -> variance x) sample   ]

file 'Boundary.hs'

...

{-# LANGUAGE FlexibleContexts #-} module Boundary where import qualified Data.Vector.Generic as G

variance :: (G.Vector v Double)          => v Double -> Double variance vec = G.sum vec {-# INLINE variance #-}

Here is benchmarking results:

benchmarking eta    - mean: 220.8042 us benchmarking lambda - mean: 24.31309 us

If variance is moved to the test.hs file or eta reduced or written as lambda: varance = \vec -> G.sum vec difference goes away. What causes such behavior?

_______________________________________________ Glasgow-haskell-users mailing list Glasgow-haskell-users@haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-users