[GHC] #8279: bad alignment in code gen yields substantial perf issue

[GHC]

#8279: bad alignment in code gen  yields substantial perf issue
--------------------------------------------+------------------------------
        Reporter:  carter                   |            Owner:
            Type:  bug                      |           Status:  new
        Priority:  highest                  |        Milestone:
       Component:  Compiler                 |          Version:  7.7
      Resolution:                           |         Keywords:
Operating System:  Unknown/Multiple         |     Architecture:
 Type of failure:  Runtime performance bug  |  Unknown/Multiple
       Test Case:                           |       Difficulty:  Unknown
        Blocking:                           |       Blocked By:
                                            |  Related Tickets:  #8082
--------------------------------------------+------------------------------

Comment (by simonmar):

 So the problem with 16-byte aligning branch targets is that many of our
 code blocks have a 3-word info table.  We would have to pad these info
 tables by one word in addition to aligning to 16 bytes.  That might not be
 too bad, but someone needs to do the measurements to see what the code
 size / speed tradeoff is.

 Also we don't necessarily want to align all our labels, because many of
 them are just heap-check failure targets and wouldn't benefit from
 aligning at all.

 I tend to optimise for small binary sizes because I was brought up on
 computers with 32K of memory and I think you should never waste a byte :-)
 If you think binary size can be won elsewhere, please do it :-P

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/8279#comment:11>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler