#2002: problems with very large (list) literals
------------------------------------------+---------------------------------
Reporter: Isaac Dupree | Owner: simonmar
Type: compile-time performance bug | Status: new
Priority: high | Milestone: 6.10 branch
Component: Compiler | Version: 6.8.2
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: x86
Os: Linux |
------------------------------------------+---------------------------------
Comment (by Isaac Dupree):
okay, I tried it today with -O [*] (compiler: 6.9.20080619, a stage1
compiled by 6.8.2) and it succeeded after two hours, using up to 738 MB
RAM. So it's "only" 40 times slower optimizing than the already-slow not-
optimizing case. [*] First I tried compiling all the modules of GHC that
Lexer.hs depends on, with -O0, then compiling Lexer.hs with -O; then I
tried again compiling the dependent modules with -O too, and the compile
time (not including dependent modules) of Lexer.hs was only a few minutes
longer with a few more megabytes used.
I avoided ghc --make for testing because that might have an additional
risk of wasted memory that I'm not trying to test.
P.S. is there a way to get a correct build order other than string-
processing on the output of ghc --make? I tried ghc -M but that gave a
makefile format and the modules weren't textually in the necessary order.
Anyway, I'm attaching a test-case extracted from the GHC source but hacked
so it doesn't need to go through the configure process (since it suffers
the same slowness being compiled).. use e.g.
{{{
./build_deps path_to_ghc -O
./build_lexer path_to_ghc -O
}}}
oh wait, trac won't let me attach that big a file (half a megabyte), so
here it is: http://isaac.cedarswampstudios.org/2008/ghc_x_slowness.tar.gz
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/2002#comment:13>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
