[GHC] #1600: Optimisation: CPR the results of IO

[GHC]

#1600: Optimisation: CPR the results of IO
-------------------------------------+-------------------------------------
        Reporter:  simonmar          |                Owner:  (none)
            Type:  task              |               Status:  new
        Priority:  lowest            |            Milestone:
       Component:  Compiler          |              Version:  6.6.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Runtime           |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #8598             |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by akio):

 It seems like the above results were bogus because it used libraries
 compiled with the wrong compiler (I had `stage=2` in build.mk. I removed
 it and typed `make`. Is it not sufficient to cause a full build?).

 After a clean build, the allocation numbers look very good:

 {{{
 --------------------------------------------------------------------------------
         Program           Size    Allocs   Runtime   Elapsed  TotalMem
 --------------------------------------------------------------------------------
       compress2          +2.0%     -0.9%     0.170     0.170     -3.7%
            fft2          +1.0%     -0.2%     0.078     0.078      0.0%
           fluid          +2.1%     -0.4%     0.006     0.006      0.0%
          gamteb          +0.4%     -4.5%     0.032     0.032      0.0%
             gcd          +0.7%    -21.4%     0.046     0.046      0.0%
           infer          -0.1%     -1.2%     0.054     0.054      0.0%
         integer          +1.0%     -1.5%    -15.0%    -15.0%      0.0%
          mandel          +1.0%    -24.4%     0.051     0.051      0.0%
            rfib          +1.0%     -0.3%     0.012     0.012      0.0%
           solid          +0.4%     -6.6%     0.130     0.131      0.0%
 --------------------------------------------------------------------------------
             Min          -0.2%    -24.4%    -21.1%    -21.3%     -3.7%
             Max          +2.1%     +0.1%    +39.3%    +39.3%    +17.3%
  Geometric Mean          +0.2%     -0.7%     -4.9%     -4.9%     +0.1%
 }}}

 Should I worry about the increase in the code size? If not I'll clean up
 the branch and submit a Diff.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/1600#comment:60>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler