[Haskell] Edison 1.2 and profiling

Mario Blazevic mblazevic at stilo.com
Sun Jun 4 22:04:10 EDT 2006

> > 
> > Enabling profiling can significantly change the space behavior of  
> > programs.  There may be some important optimizations that don't occur  
> > in the presence of profiling, and the profiling data itself takes  
> > some memory.  I have to admit, I don't know a lot about GHC profiling  
> > so I'm afraid I can't be of much help.
> Me neither. I tried several things: Just turning off the optimizations doesn't cause the
> problem. Optimization + profiling (-O -prof) has the same problem as -prof alone. It
> doesn't matter if -auto-all option is there or not. And the profiled program gobbles up
> all memory with or without the +RTS -p option.

Furthermore, the problem has nothing to do with Edison. The version 6.4.2 of ghc is the
culprit. I hadn't run the profiler since I upgraded to 6.4.2 until now. But now when I
compile with 6.4.2 the old version of the program, using Data.Map, I see the same behavior
as with PatriciaLoMap: the program diverges and eats all available memory.

I thought it could be the GCC 4.1.1 which was also a recent upgrade, but the behavior is
the same with -prof -fasm options. So I guess it must be ghc. Did anybody else encounter
this problem?

More information about the Haskell mailing list