#1434: Slow conversion from Double to Int
---------------------------------------+------------------------------------
Reporter: [EMAIL PROTECTED] | Owner: dons
Type: bug | Status: new
Priority: normal | Milestone: 6.8 branch
Component: libraries/base | Version: 6.4.1
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: Unknown
Os: Linux |
---------------------------------------+------------------------------------
Changes (by dons):
* owner: => dons
* cc: [EMAIL PROTECTED] (added)
Comment:
I looked into this after the realToFrac issues last week. It's a similar
story -- the noop rules are currently only set up for Int conversions.
So, for example, this program:
{{{
import Data.Word
import Data.Array.Vector
main = print . sumU
. mapU (truncate::Double->Word16)
$ (enumFromToFracU 0.1 (100000000 :: Double))
}}}
truncate stays as a call to properFraction.
Now, we can add some rules for this stuff:
{{{
{-# RULES
"truncate/Double->Int64"
truncate = fromIntegral . double2Int :: Double -> Int64
"truncate/Double->Int32"
truncate = fromIntegral . double2Int :: Double -> Int32
"truncate/Double->Int16"
truncate = fromIntegral . double2Int :: Double -> Int16
"truncate/Double->Int8"
truncate = fromIntegral . double2Int :: Double -> Int8
"truncate/Double->Word64"
truncate = fromIntegral . double2Int :: Double -> Word64
"truncate/Double->Word32"
truncate = fromIntegral . double2Int :: Double -> Word32
"truncate/Double->Word16"
truncate = fromIntegral . double2Int :: Double -> Word16
"truncate/Double->Word8"
truncate = fromIntegral . double2Int :: Double -> Word8
#-}
}}}
And the running time drops from:
{{{
$ time ./henning
28800
./henning 45.84s user 0.11s system 98% cpu 46.448 total
}}}
before, to:
{{{
$ time ./henning
28800
./henning 0.39s user 0.00s system 97% cpu 0.398 total
}}}
after.
I think this is as good a time as any to do an audit of the Int/noop
conversion rules
in base, looking for others that also should work for integral types. I'll
take care of this.
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/1434#comment:6>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
