[Haskell-beginners] Performance of Idiomatic lazy Haskell

[Markus Böhm]

Daniel, here another feedback:

F:\MeineUebungen>ghc -O2 -fexcess-precision -fvia-C -optc-O3 -o luaLoop --make p
085-pi_lualoop.hs
[1 of 1] Compiling Main             ( p085-pi_lualoop.hs, p085-pi_lualoop.o )
Linking luaLoop.exe ...

F:\MeineUebungen>echo '0.00000001' | time ./luaLoop
Eingegebene Zeit kann nicht übernommen werden.
Geben Sie die neue Zeit ein:

F:\MeineUebungen>luaLoop +RTS -sstderr -RTS
luaLoop +RTS -sstderr
EPS:
0.00000001
3.1415926485894725
          61,428 bytes allocated in the heap
           1,316 bytes copied during GC
           4,564 bytes maximum residency (1 sample(s))
          11,820 bytes maximum slop
               1 MB total memory in use (0 MB lost due to fragmentation)

  Generation 0:     0 collections,     0 parallel,  0.00s,  0.00s elapsed
  Generation 1:     1 collections,     0 parallel,  0.00s,  0.00s elapsed

  INIT  time    0.02s  (  0.00s elapsed)
  MUT   time    8.06s  ( 16.83s elapsed)
  GC    time    0.00s  (  0.00s elapsed)
  EXIT  time    0.00s  (  0.00s elapsed)
  Total time    8.08s  ( 16.83s elapsed)

  %GC time       0.0%  (0.0% elapsed)

  Alloc rate    7,604 bytes per MUT second

  Productivity  99.8% of total user, 47.9% of total elapsed

-- Markus




On Mon, Feb 1, 2010 at 3:08 PM, Markus Böhm <markus.boehm at googlemail.com> wrote:
> Daniel, I use GHC 6.12.1 and Windows XP. The time command doesn't seem
> to work. It says in German: specified time can't be read. Give a new
> time.
>
> -- Markus
>
>
>
>
> On Mon, Feb 1, 2010 at 2:52 PM, Daniel Fischer <daniel.is.fischer at web.de> wrote:
>> Am Montag 01 Februar 2010 14:31:40 schrieb Markus Böhm:
>>> Daniel, with LuaJIT it needs 1.45 sec cpu-time on my machine with
>>> attached variant.
>>
>> That's pretty fast.
>>
>>> I compiled all our Haskell variants with ghc --make -O2.
>>
>> Well, GHC isn't as good at loop-optimising as gcc is. Depending on the
>> loop, the via-C compiled binaries are between 1.4 and 2.3 times faster than
>> the NCG compiled.
>>
>>> I don't know about -fvia-C, have to find out. I hope I didn't
>>> distract You with my Lua variant.
>>
>> No sweat.
>>
>>> In any case thank You very much for
>>> your advice. Markus.
>>>
>>
>> You're welcome.
>>
>> Can you try the below with
>>
>> ghc -O2 -fexcess-precision -fvia-C -optc-O3 -o luaLoop --make Whatever.hs
>>
>> and run with
>>
>> echo '0.00000001' | time ./luaLoop
>>
>> ?
>> It's a fairly direct translation of the Lua code, and it runs here more or
>> less equally fast as (gcc compiled) C-loops.
>>
>> =================================
>> module Main (main) where
>>
>> main :: IO ()
>> main = do
>>    putStrLn "EPS:"
>>    eps <- readLn :: IO Double
>>    print $ 4*calcPi eps
>>
>> calcPi :: Double -> Double
>> calcPi eps = go False 1 3
>>      where
>>        go bl p1 i
>>            | 4*abs(p2-p1) < eps    = p1
>>            | otherwise             = go (not bl) p2 (i+2)
>>              where
>>                p2  | bl        = p1+1/i
>>                    | otherwise = p1-1/i
>>
>> ==================================
>>
>>
>