[GHC] #8027: Adding one call to getNumCapabilities triggers performance nose dive (6X slowdown)

[GHC]

#8027: Adding one call to getNumCapabilities triggers performance nose dive (6X
slowdown)
--------------------------------------+-------------------------------------
  Reporter:  rrnewton                 |          Owner:                  
      Type:  bug                      |         Status:  closed          
  Priority:  normal                   |      Milestone:                  
 Component:  Runtime System           |        Version:  7.6.3           
Resolution:  wontfix                  |       Keywords:                  
        Os:  Unknown/Multiple         |   Architecture:  Unknown/Multiple
   Failure:  Runtime performance bug  |     Difficulty:  Unknown         
  Testcase:                           |      Blockedby:                  
  Blocking:                           |        Related:                  
--------------------------------------+-------------------------------------
Changes (by rrnewton):

  * status:  new => closed
  * resolution:  => wontfix


Comment:

 Ah... my mistake.  What was so apparently baffling is actually pretty
 reasonable.

 In this case adding more code to the function simply changes inlining
 decisions, which in turn determines whether dictionary elimination can be
 done.  The 6X performance difference is just because of that.

 Perhaps such a function should be monomorphic to be safe... since the
 difference is so big.

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/8027#comment:6>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler