Fortran - Fortran 90 compiler on Mac

Author

Fortran 90 compiler on Mac

guzik@astro.upenn.edu

2006-01-18, 7:03 pm

Hello All,

I'm looking for a free Fortran 90 compiler for a PowerBook which works
comparably
nice as Intel ifort for PC. I gave gfortran from gcc-4 a try but it
fails to compile Numerical Recipes
(Press et al) modules.

I'm sure there are people around who solved the problem with
compilation
of NR modules. Could you please share your knowledge with me?
Also, maybe you could point me an another free compiler for Mac.

Thanks a lot,
Jacek

Rich Townsend

2006-01-18, 7:03 pm

guzik@astro.upenn.edu wrote:
> Hello All,
>
> I'm looking for a free Fortran 90 compiler for a PowerBook which works
> comparably
> nice as Intel ifort for PC. I gave gfortran from gcc-4 a try but it
> fails to compile Numerical Recipes
> (Press et al) modules.

Could you show us the error messages? And have you reported the problem to the
gfortran development team?

>
> I'm sure there are people around who solved the problem with
> compilation
> of NR modules. Could you please share your knowledge with me?
> Also, maybe you could point me an another free compiler for Mac.

You may want to give g95 a shot -- it seems to be rather further along the
development cycle than gfortran.

cheers,

Rich

Christopher Hulbert

2006-01-18, 7:03 pm

guzik@astro.upenn.edu wrote:
> Hello All,
>
> I'm looking for a free Fortran 90 compiler for a PowerBook which works
> comparably
> nice as Intel ifort for PC. I gave gfortran from gcc-4 a try but it
> fails to compile Numerical Recipes
> (Press et al) modules.
>
> I'm sure there are people around who solved the problem with
> compilation
> of NR modules. Could you please share your knowledge with me?
> Also, maybe you could point me an another free compiler for Mac.
>
> Thanks a lot,
> Jacek
>

Have you tried g95? It lists Numerical Recipes as one of the compiled software
in the status.

http://www.g95.org

Richard E Maine

2006-01-18, 7:03 pm

<guzik@astro.upenn.edu> wrote:

> I'm looking for a free Fortran 90 compiler for a PowerBook which works
> comparably nice as Intel ifort for PC.

I don't know what "works comparably nice" means. It could have various
interpretations, but I assume from the following that you are talking
about being able to correctly compile code, rather than something about
the interface.

> I gave gfortran from gcc-4 a try but it
> fails to compile Numerical Recipes...

While I haven't tried Numerical Recipes, my general experience has been
that g95 (<http://www.g95.org> ) compiles substantially more of my codes
than gfortran does.

The last few time I tried, gfortran couldn't compile anything on my Mac
at work apparently because of some missing library. I confess I didn't
take the time to track the problem down properly, and I'll grant it
might be just some quirk of my particular sytem configuration. But even
when I did have a working installation of gfortran on my Mac, it lagged
seriously behind g95 in terms of compiling my codes. (And no, I do not
consider the probleems to have been in arcane areas of F90).

Gfortran does seem to be improving, but if you are having problems with
it, g95 is another option. Some people confuse g95 and gfortran because
of their related roots, but they are not the same compiler.

If you need all three of f90, free, and Mac, those two compilers are
pretty much the only two options. There are other options if you give up
on any of the three parts.

--
Richard Maine | Good judgment comes from experience;
email: my first.last at org.domain| experience comes from bad judgment.
org: nasa, domain: gov | -- Mark Twain

Steven G. Kargl

2006-01-18, 7:03 pm

In article <dqm1cj$1pq$1@scrotar.nss.udel.edu>,
Rich Townsend <rhdt@barVOIDtol.udel.edu> writes:
> guzik@astro.upenn.edu wrote:
>
> Could you show us the error messages? And have you reported the
> problem to the gfortran development team?
>

There is a known problem with compiling NR. I know what needs to
be done to fix it, but I have neither the time nor a sufficiently
complete knowledge of all the strange ways one may abuse TRANSFER.

--
Steve
http://troutmask.apl.washington.edu/~kargl/

Jacek

2006-01-18, 7:03 pm

Thanks a lot for your help.

It seems that the problem I encounter is a known bug in gfortran.

I'll definitely have to give g95 a try.

Thanks,
Jacek

max

2006-01-18, 7:03 pm

<guzik@astro.upenn.edu> wrote in message
news:1137608459.669311.292250@z14g2000cwz.googlegroups.com...
>
> Hello All,
>
> I'm looking for a free Fortran 90 compiler for a PowerBook which works
> comparably
> nice as Intel ifort for PC. I gave gfortran from gcc-4 a try but it
> fails to compile Numerical Recipes
> (Press et al) modules.
>
> I'm sure there are people around who solved the problem with
> compilation
> of NR modules. Could you please share your knowledge with me?
> Also, maybe you could point me an another free compiler for Mac.
>
> Thanks a lot,
> Jacek
>

If you're willing to pay, then try Absoft Fortran. Your backstop if gfortran
& g95 fail. Or switch to using the Fortran 77 versions of NR. Better yet,
look at Netlib for alternatives.

Richard E Maine

2006-01-18, 7:03 pm

max <none@none.com> wrote:

> If you're willing to pay, then try Absoft Fortran....

Or if you are willing to pay, but not quite so much :-), there is, of
course NAG's compiler. The Absoft one is $900 commercial, or $675
educational, either of which is just plain out of the question for lots
of people at home (work use is another matter). Nag's commercial price
is about $600, and educational $240. The factor of nearly 3 difference
in edu prices is substantial in my book.

I presume both to be quality products. I use the NAG compiler regularly.
I've never actually tried Absoft because the odds of me putting out that
much money for a copy of it at home are zero, even with the edu rate,
and we already have other compilers at work.

--
Richard Maine | Good judgment comes from experience;
email: my first.last at org.domain| experience comes from bad judgment.
org: nasa, domain: gov | -- Mark Twain

Gordon Sande

2006-01-18, 10:00 pm

On 2006-01-18 19:44:41 -0400, nospam@see.signature (Richard E Maine) said:

> max <none@none.com> wrote:
>
>
> Or if you are willing to pay, but not quite so much :-), there is, of
> course NAG's compiler. The Absoft one is $900 commercial, or $675
> educational, either of which is just plain out of the question for lots
> of people at home (work use is another matter). Nag's commercial price
> is about $600, and educational $240. The factor of nearly 3 difference
> in edu prices is substantial in my book.
>
> I presume both to be quality products. I use the NAG compiler regularly.
> I've never actually tried Absoft because the odds of me putting out that
> much money for a copy of it at home are zero, even with the edu rate,
> and we already have other compilers at work.

A feature of the NAG compiler is its debugging. It has the usual subscript
stuff but also has undefined variable checking as of V5. They had V5 on
the Mac before it was on some other platforms.

If you have never used undefined variable checking you may think it
is yet another check mark feature. But once you figure out how useful
it is you tend to make it a deal breaking feature. If the super best
optimizing compiler is important then you will (almost certainly) have
no trouble justifying one copy of every compiler but otherwise you
should go with the best debugging features.

Sponsored Links