[GHC] #15418: Performance drop 60 times on non-profiling binary
GHC
ghc-devs at haskell.org
Wed Aug 29 19:55:04 UTC 2018
#15418: Performance drop 60 times on non-profiling binary
-------------------------------------+-------------------------------------
Reporter: hth313 | Owner: (none)
Type: bug | Status: infoneeded
Priority: high | Milestone: 8.8.1
Component: Runtime System | Version: 8.4.3
Resolution: | Keywords:
Operating System: MacOS X | Architecture: x86_64
Type of failure: Runtime | (amd64)
performance bug | Test Case:
Blocked By: | Blocking:
Related Tickets: #14414, #9599 | Differential Rev(s):
Wiki Page: |
-------------------------------------+-------------------------------------
Comment (by osa1):
To be honest I don't think that'd be helpful. Normally for this I'd start
with
comparing Core outputs of the program, with and without profiling flags. I
think it's very likely that this will be enough to see why profiled
version is
faster.
Btw, looking at the previous comments I realized that we didn't check
different
optimisation parameters. One of the reasons why sometimes profiling builds
are
faster is because profiling sometimes inhibits some optimisations (as it
changes Core of the program), and optimisations sometimes make some
programs
slower. Could you try building your program without profiling, and with
different optimisation settings? Please report numbers for:
- -O0
- -O1 (default when using cabal)
- -O2
If you're using cabal add these to `ghc-options` in your .cabal file.
Depending
on the numbers you can then try to disable individual optimisation passes
(see
the user manual or man page) etc.
--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15418#comment:17>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
More information about the ghc-tickets
mailing list