[GHC] #14981: GHC parallel GC is not doing well on modern many-core machine

GHC ghc-devs at haskell.org
Tue Mar 27 20:13:25 UTC 2018


#14981: GHC parallel GC is not doing well on modern many-core machine
-------------------------------------+-------------------------------------
        Reporter:  varosi            |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Runtime System    |              Version:  8.4.1
      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:                    |
-------------------------------------+-------------------------------------

Comment (by bgamari):

 Minor GCs indeed do necessarily stop-the-world. This is a known limitation
 which affects large core-counts particularly badly.

 However, fixing this in a copying garbage collector is quite tricky. There
 have been a few attempts at avoiding this stop-the-world. The most recent
 attempt is the Simons' "Multicore Garbage Collection with Local Heaps".
 You can still find the prototype implementation (against GHC 6.10, IIRC)
 on the `wip/local-heaps` branch but it not merged as the performance
 improvement of brought by this change was outweighed by its enormous
 complexity.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14981#comment:6>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler


More information about the ghc-tickets mailing list