[GHC] #14565: Performance degrades from -O1 to -O2 (was: Memory leak switching from -O1 to -O2)

GHC ghc-devs at haskell.org
Fri Dec 8 14:09:05 UTC 2017


#14565: Performance degrades from -O1 to -O2
-------------------------------------+-------------------------------------
        Reporter:  dbeacham          |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.2.2
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:  x86_64
 Type of failure:  Runtime           |  (amd64)
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:  #14379            |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------
Description changed by dbeacham:

Old description:

> GHC seems to get stuck at SpecConstr stage when compiling. Cpu and memory
> usage then sky rocket.
>
> I've managed to get the example down to what appears to be minimal but
> making a number of minor changes makes the compilation go through
> quickly:
>
> * changing `toIdx` to `const 0`
> * removing one of the `V.forM_` layers.
> * using "-fno-spec-constr" (unsuprisingly).
>
> I can't reproduce it on 8.0.2.

New description:

 GHC seems to get stuck at SpecConstr stage when compiling.

 I've managed to get the example down to what appears to be minimal but
 making a number of minor changes makes the compilation go through quickly:

 * changing `toIdx` to `const 0`
 * removing one of the `V.forM_` layers.
 * using "-fno-spec-constr" (unsuprisingly).

 I can't reproduce it on 8.0.2.


 In larger examples I was able to get memory and cpu to sky rocket quite
 quickly but I appear to have lost some of that behaviour when paring down
 the example.

--

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14565#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler


More information about the ghc-tickets mailing list