[GHC] #15938: Hadrian's recompilation check is extremely slow

GHC ghc-devs at haskell.org
Thu Nov 29 22:55:29 UTC 2018


#15938: Hadrian's recompilation check is extremely slow
-------------------------------------+-------------------------------------
        Reporter:  bgamari           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  high              |            Milestone:  8.8.1
       Component:  Build System      |              Version:  8.6.2
  (Hadrian)                          |
      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):

 Looking at the eventlog under threadscope suggests that we are hitting
 some very peculiar GC behavior. Hadrian is linked against the threaded,
 profiled, eventlog-enabled RTS but with `+RTS -N1`. From t=1 until around
 t=22 seconds we have very frequent GCs, with time roughly roughly evenly
 between major and minor collections. The productivity during this period
 is roughly 20%.

 Intriguingly, increasing the allocation area size to 30MB reduces the
 start-up GC time to around 1.5 seconds, with mutator time remaining around
 12 seconds (a productivity of roughly 90%).

 By the way, I am bootstrapping and building Hadrian with GHC 8.4.3.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15938#comment:13>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler


More information about the ghc-tickets mailing list