[GHC] #9221: (super!) linear slowdown of parallel builds on 40 core machine

[GHC]

#9221: (super!) linear slowdown of parallel builds on 40 core machine
-------------------------------------+-------------------------------------
        Reporter:  carter            |                Owner:
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:  8.2.1
       Component:  Compiler          |              Version:  7.8.2
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #910, #8224       |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by carter):

 Very cool work!

 Do we want to limit the number of capabilities to the number of hyper
 threaded cores or the number of physical cores?

 Likewise, another opportunity for improvement might be seeing about making
 sure the capabilities (when CPU pining is enabled for the rts) are spread
 out maximally across the available CPU cores?  I believe this or a similar
 issue was discussed / observed by Ryan Yates / fryguybob previously but
 I'm not sure if we had worked out a portable way to decide this or not

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/9221#comment:60>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler