[GHC] #11054: GHC on Windows could not use more than 64 logical processors

[GHC]

#11054: GHC on Windows could not use more than 64 logical processors
-------------------------------------+-------------------------------------
        Reporter:  varosi            |                Owner:
            Type:  feature request   |               Status:  closed
        Priority:  normal            |            Milestone:  8.2.1
       Component:  Runtime System    |              Version:  7.10.2
      Resolution:  fixed             |             Keywords:
Operating System:  Windows           |         Architecture:  x86_64
 Type of failure:  Runtime           |  (amd64)
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #12602            |  Differential Rev(s):  Phab:D2533
       Wiki Page:                    |
-------------------------------------+-------------------------------------
Changes (by Phyx-):

 * status:  new => closed
 * resolution:   => fixed


Comment:

 I have been looking this code over, and the failures shouldn't affect the
 performance since the RTS constantly changes affinity. So the fail would
 only affect a tiny slice of time.

 I'm not sure why from the hundreds of events only one or two fails, but
 double checking the code by hand I can't see a problem with the request,
 which leads me to believe the OS must be rejecting it for another reason
 than an invalid mask.

 Even with one processor group the example doesn't pass 90-93% on my
 machine. Which leads me to believe this is a general issue with the RTS
 and not this code. I think maybe because of the artificial nature of the
 testcase that the RTS might be doing a lot of GC, and that's eating time.

 If you want to know for sure please mail the mailing list.

 Because this code is correct, I will close this ticket as the logs show it
 being able to set the affinity over each of the cores correctly in all
 groups.

 About the defaults, you should really bring it up with the community if
 you feel that strongly about it. I am not really in a position to comment
 about what is sensible or not in this case.

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11054#comment:47>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler