Re: [GHC] #16122: `round :: Double -> Int64` much slower than `fromIntegral @Int @Int64 . round`

4 Jan 2019

      #16122: `round :: Double -> Int64` much slower than `fromIntegral @Int @Int64 .
round`
-------------------------------------+-------------------------------------
        Reporter:  Fuuzetsu          |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.6.3
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------
Changes (by akio):

 * cc: akio (added)

Comment:

 In the interface file for GHC.Int I see:

 {{{
 "round/Double->Int64" forall ($dRealFrac :: GHC.Real.RealFrac
                                               GHC.Types.Double)
                              ($dIntegral :: GHC.Real.Integral Int64)
   GHC.Real.round @ GHC.Types.Double $dRealFrac @ Int64 $dIntegral
   = GHC.Base..
       @ GHC.Types.Int
       @ Int64
       @ GHC.Types.Double
       (GHC.Real.fromIntegral
          @ GHC.Types.Int
          @ Int64
          GHC.Real.$fIntegralInt
          $fNumInt64)
       (GHC.Float.$fRealFracDouble_$cround
          @ GHC.Types.Int
          GHC.Real.$fIntegralInt)
 }}}

 So the RHS of the RULE already contains the bad code.

 This appears to be caused by the fact that the RHS of a rule is rewritten
 with GHC's built-in rules, even when there is a conflicting user-defined
 one.

-- 
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/16122#comment:3
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler