On 9 Mar 2002, Bryan Paxton wrote:

> >  It perhaps might be the latest gcc, specifically libstdc++, but it
> > could be my optimizations that made something in libstdc++ fubar...
> I'm
> > going to compile for i686, but with the default rpm opt flags and see
> > what's what : ) 
> > 
> >  I'll email back with results : )
> 
> Those being that either libstdc doesn't like compiling with the below
> flags, or gcc is happy with them (2.9.6 from cooker):
> i686 -03 -funroll-loops -ffast-math -malign-double -mcpu=pentiumpro
> -march=pentiumpro -fomit-frame-pointer -fno-exceptions
> -fno-strength-reduce

I've encountered quite a few 'risky' programs that segfaults with
too much optimization, turning off -ffast-math and/or
-fno-strength-reduce usually helps. Just my experience.

Abel


> 
> (PIII - 850)
> 
> Anyway, so no problem with libstdc++ : )
> 
> 



Reply via email to