[GHC] #16040: Unboxing-Related Performance Issue with Polymorphic Functions

[GHC]

#16040: Unboxing-Related Performance Issue with Polymorphic Functions
-------------------------------------+-------------------------------------
        Reporter:  _recursion        |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.6.2
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Runtime           |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by _recursion):

 Thank you for the detailed explanation, Simon! I took a look at the core
 but missed seeing the allocation previously.

 Regarding fixing this or CPR, would nested CPR fix all cases 'like this',
 as you put it, or even if the nested CPR patch lands would there still be
 a need for functionality to catch cases like this?

 The reason I ask is that I'd be quite willing to take a look at either
 over the holiday in some free time, assuming I can find some guidance, and
 I'm wondering which would be best to attack.

 Regarding the nested-cpr patch, what kind of attention is needed? Do you
 think that, ''with'' said attention it could be landed for 8.8?

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/16040#comment:6>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler