Hi Ondrej,
On Dec 24, 10:37 pm, "Ondrej Certik" <[EMAIL PROTECTED]> wrote: > > I am quite shocked that g77 is that far ahead on the performance > > curve. I am sure I would have heard about it by now if that was the > > general case, but is there any chance your code might hit some corner > > case in gfortran? Which gfortran did you use exactly? Does using g95 > > $ gfortran --version > GNU Fortran (GCC) 4.2.3 20071123 (prerelease) (Debian 4.2.2-4) > Copyright (C) 2007 Free Software Foundation, Inc. Ok, that seems reasonably current ;) > > make a difference? > > As discussed in the IRC session attached in the previous email, the > pure difference is just 1.0 (g77) vs 1.3s (gfortran), so not that big. > One problem was in the random() function, being fast in glibc, slower > in g77 and the slowest in gfortran. Ok, I over read that detail, the mail was pretty long, but thanks for the clarification. > I didn't try g95, as it is not in Debian. I believe in standard tools > and "official" ways of doing things, > which currently is gfortran in Debian. Of course, if there are some > very good reasons, not to follow the > same way as the majority of other people, we can do that, i.e. package > g95 for Debian, recompile > debian packages with g95, instead of gfortran (currently Debian uses > g77, but it's moving to gfortran), > find a way how to ship both binary packages in Debian (compiled with > both gfortran and g95) etc. > > But I am not really looking into speeding up this particular code, but > just finding ways what people > use and are going to use. Fortran will definitely not die (at least > not soon), I consider that unlikely to happen, but I certainly won't be sad when the use of Fortran were to decline over the next couple years. > but I still feel that C > is much more widespread language, with comparable speed. > > Ondrej Cheers, Michael --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~----------~----~----~----~------~----~------~--~---
