g77, g95, optimization

[Steve Kargl]

On Mon, Jun 18, 2007 at 11:29:40PM +0200, Terry Frankcombe wrote:
> An aside...
> 
> On Mon, 2007-06-18 at 14:10 -0700, Steve Kargl wrote:
> > > 3) I have a PC with an Intel Core 2 Duo E6700 CPU which runs at 2.66
> > > Ghz.
> > 
> > With gfortran, try "-march=native -O3 -ffast-math -ftree-vectorize
> > -funroll-loops"  I don't normally recommend -ffast-math, but you
> > asked for the fastest execution (without specifying if you wanted
> > correct results. :-)
> 
> I humbly submit Steve's email as evidence for my assertion that
> gfortran/gcc dislikes simple flags to obtain simple results.  -fast,
> anyone?  ;-)
> 
> <http://gcc.gnu.org/ml/fortran/2007-05/msg00301.html>
> <http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31114>
> 
> ...and now Tobias has given another, different(!) long set of flags...

Actually, our sets of options are quite similar.  Tobias
added -msse3, which I thought was turned on be default for
Intel Core 2 Duo.

My -march=native is equilavent to Tobias' -march=nocona.  The 
difference is that I never can remember all of the odd codenames
that AMD and Intel choose for their cpus.  Additionally, my Makefiles
with -march=native work on any of my systems without editing.

-- 
Steve