MKL and Intel compilers were as good as AMD supplied ones on AMD machines upto some version (10.x possibly I don't remember exactly). After that version, AMD performance started to degrade.
Also, I would like to add if I may, obtaining a working executable does not necessarily mean that the executable will output what you expect. Actually, the "optimization methods" compilers employ make a significant difference. In my experience the IBM compilers are the most problematic, since I get large numerical drifts, (on the level 1e-8 per operation, although I am using double precision) and the code I am working on is especially vulnerable to those. I vaguely remember having to change how I write some loops due to how loop unrolling is handled in gfortran. In any case, please test your outcome. In case you get unphysical results, please mention it here in the forum so that we can have a look. At least people will know. Best, Baris 2010/3/16 Paolo Giannozzi <giannozz at democritos.it> > On Mar 16, 2010, at 20:31 , Carlo Nervi wrote: > > > You mentioned that gfortran is buggy. Which version is buggy? > > it depends. Recent versions typically produce a working executable > for pw.x , but not yet a working phonon executable. All I know about > the various compilers is written in the user guide (and continuously > updated in the cvs version) > > P. > --- > Paolo Giannozzi, Dept of Physics, University of Udine > via delle Scienze 208, 33100 Udine, Italy > Phone +39-0432-558216, fax +39-0432-558222 > > > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.democritos.it/pipermail/pw_forum/attachments/20100317/d47801f5/attachment.htm
