[GHC] #8199: Get rid of HEAP_ALLOCED

[GHC]

#8199: Get rid of HEAP_ALLOCED
-------------------------------------+-------------------------------------
              Reporter:  ezyang      |            Owner:  ezyang
                  Type:  feature     |           Status:  new
  request                            |        Milestone:  7.10.1
              Priority:  normal      |          Version:  7.7
             Component:  Compiler    |         Keywords:
            Resolution:              |     Architecture:  Unknown/Multiple
      Operating System:              |       Difficulty:  Project (more
  Unknown/Multiple                   |  than a week)
       Type of failure:              |       Blocked By:  5435
  None/Unknown                       |  Related Tickets:
             Test Case:              |
              Blocking:              |
Differential Revisions:  D207        |
-------------------------------------+-------------------------------------

Comment (by ezyang):

 So, here's the status: I rebased the patches onto HEAD, and took a close
 look at the oscillating results. As it turns out, at least for
 cryptarithm, the nofib test had bimodal output behavior, so depending on
 your luck, you'd end up with even changes, or an improvement or a
 regression. (I think this may be what's happening for the inconsistent GC
 results; more investigation necessary.) The increase in total memory for
 cryptarithm1 and circsim stems from the fact that the binaries don't use
 much memory to begin with, so the allocation of static closures is causing
 us to allocate a new megablock and, e.g., doubling it in the case
 cryptarithm.

 The startup time for GHC is still 0.020, so same as before.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8199#comment:37>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler