[GHC] #13570: CoreFVs patch makes n-body slower

[GHC]

#13570: CoreFVs patch makes n-body slower
-------------------------------------+-------------------------------------
        Reporter:  simonpj           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.0.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------
Changes (by simonpj):

 * cc: kavon@… (added)


Comment:

 I'm a bit surprised that the Sinking pass in Cmm didn't convert 1,2,3 into
 2,1,3.  In which case both programs would have the same end product.

 So you've at least confirmed that the runtime change was indeed (as I
 thought) something v subtle to do with prefetching or whatnot. It looks
 unreasonable for the `FloatIn` pass to choose the "right" thing, even if
 we knew what was "right".

 So it's still interesting, but it removes the urgency of saying "do we
 want this particular patch".

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13570#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler