Posted by Valera on February 02, 2007 at 16:19:20:
In Reply to: Re: Compilation with gfortran posted by Jellby on February 02, 2007 at 10:52:53:
"Out of box" solution for gf.conf will be available in Molcas 7.
: On the other hand, by turning on -funsafe-math-optimizations (implicit in -ffast-math), the resulting test calculation hangs forever.
There is no problem with -ffast-math in the current snapshot version of Molcas
: - Additionaly, the file src/util/pack.f contains the line "DATA IONES/ Z'FFFFFFFF' /"
this is 'dead' and unused code, should be removed from distribution.
: With the above changes (and -march=athlon-xp) I get all tests run (molcas verify) in 18:40 minutes, whereas with g77 it's 22:20 (both configured with -speed fast).
I must confess that my statement about the speed of molcas with gfortran was based on experiments I made with GCC 4.0.x.
The performance with 4.1.1 is much better. However, I still disagree, that gfortran always overperformed g77, - I can say, that now they are quite close.
Seward (which is obviously run in all tests) indeed, runs faster with gfortran. But RASSCF doesn't.
To make more clear picture for performance of molcas codes - I would recommend to run:
molcas verify performance - to run our benchmark suite
molcas timing - to generate a report file