[GHC] #7428: GHC compile times are seriously non-linear in program size (was: Occurrence analyzer out of memory)

[GHC]

#7428: GHC compile times are seriously non-linear in program size
---------------------------------+------------------------------------------
  Reporter:  nudded              |          Owner:                
      Type:  bug                 |         Status:  new           
  Priority:  normal              |      Milestone:                
 Component:  Compiler            |        Version:  7.4.2         
Resolution:                      |       Keywords:                
        Os:  MacOS X             |   Architecture:  x86_64 (amd64)
   Failure:  Compile-time crash  |     Difficulty:  Unknown       
  Testcase:                      |      Blockedby:                
  Blocking:                      |        Related:                
---------------------------------+------------------------------------------
Changes (by simonpj):

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


Comment:

 I'm re-opening becuase I believe `nudded` is telling us that compilation
 times are seriously non-linear in program size. That should not happen.
 Moreover you were kind enough to profile the compiler and in the profile
 you gave it seems that 64% of all allocation is in
 `MkIface.addFingerprints` which is not supposed to be an expensive
 function.

 Could you help us some more?
  * Give a table of compile times against program size (or number of state
 transformers)
  * Explain exactly how to reproduce (what flags, what changes to make to
 the code to make N=1, N=2, N=4, etc
  * Use `-dshow-passes` to see if the size of the simplified program is
 blowing up
  * If you feel up to it, drill down into `addFingerprints` to see where
 the time is going.

 Thanks!

 Simon

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/7428#comment:7>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler