[GHC] #11443: SPECIALIZE pragma does not work + compilation times regression in GHC 8.0-rc1

[GHC]

#11443: SPECIALIZE pragma does not work + compilation times regression in GHC
8.0-rc1
-------------------------------------+-------------------------------------
        Reporter:  danilo2           |                Owner:  bgamari
            Type:  bug               |               Status:  new
        Priority:  highest           |            Milestone:  8.0.2
       Component:  Compiler          |              Version:  8.0.1-rc1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by bgamari):

 Looking at the `verbose-core2core` output from `Test2` it seems quite
 clear that the specialisation rules are firing as expected: While the
 `dump-ds` output contains an application of `perfc1' @Int ...` as
 expected, this is rewritten to `A.$sperfc3 ...` in the first simplifier
 phase (where `$sperfc3` is indeed the expected `Int`-specialised binding).

 So, the question is: why are things slowing down despite this? I know that
 Richard did make some changes in how type families are reduced (see
 3f5d1a13f112f34d992f6b74656d64d95a3f506d and
 3e1b8824c849d063c7354dbdf63ae2910cf0fdfc). Perhaps the next place to look
 is the tc-trace output.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11443#comment:15>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler