[GHC] #3034: divInt# floated into a position which leads to low arity

[GHC]

#3034: divInt# floated into a position which leads to low arity
--------------------------------------------+------------------------------
        Reporter:  batterseapower           |            Owner:
            Type:  bug                      |           Status:  patch
        Priority:  lowest                   |        Milestone:  7.6.2
       Component:  Compiler                 |          Version:  6.10.1
      Resolution:                           |         Keywords:
Operating System:  Unknown/Multiple         |     Architecture:
 Type of failure:  Runtime performance bug  |  Unknown/Multiple
       Test Case:                           |       Difficulty:  Unknown
        Blocking:                           |       Blocked By:
                                            |  Related Tickets:
--------------------------------------------+------------------------------
Changes (by nomeata):

 * status:  new => patch


Comment:

 Changing `primOpIsCheap` to always True validates, and has no allocation
 change in nofib. I guess I should be measuring runtime numbers, but for
 that I’ll need the machine to be unloaded, maybe later today....

 Alright, numbers are in:
 {{{
             Min          -0.0%     -0.0%     -7.1%     -6.5%     -1.8%
             Max          +0.0%     +0.0%     +2.1%     +1.5%    +19.8%
  Geometric Mean          -0.0%     -0.0%     -0.6%     -0.5%     +0.2%
 }}}
 so it seems to be worth it.

 Not pushing quite yet: Assuming we make the cheapness configurable in
 `primops.txt.pp`, what primops should not be cheap? To me all of them look
 cheap (e.g. arithmetic) or are side-effecting (like `copyArray#`, in which
 case the `primOpIsCheap` information hopefully has no effect. So it seems
 to me that we do not need to a flag in `primops.txt.pp` just yet; maybe
 something will show up later.

 (Marking as for review without a patch because the patch is trivial; it is
 this question that need to be reviewed.)

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/3034#comment:14>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler