[Beowulf] g77 limits...

[Robert G. Brown]

On Fri, 24 Feb 2006, Chris Samuel wrote:

> On Friday 24 February 2006 06:40, Toon Moene wrote:
>
>> The following code works on my 0.5 Gbyte RAM / 1.5 Gbyte SWAP Athlon 64
>> laptop with gfortran (the new Fortran 95 GNU Fortran)
>
> I note that everyone is talking about G95/GFortran whilst RGB mentioned that
> this code will have to run on random boxes with random distributions and I
> believe that he is asking for g77 specifically because that's the lowest
> common denominator he can guarantee.
>
> I do know you can build GCC 4.* on RH7.3 but I don't know if he can rely on
> that having been done (or mandate it for this code).
>
> Rob, is that a fair summary ?

Yup.  Although today I think I convinced one of the project PIs that
'twere better to just code things up with modern fortran (and indeed,
update the f77 code to f95 or thereabouts and not just recompile it with
a newer compiler:-) and let the RH 7.3 folks fend for themselves.  It
isn't like it isn't trivial to install an up-to-date linux distro, these
days.  Indeed, we can provide 2-3 ready-to-install distro repositories
on whatever system we use to distribute the software itself.

For sure it is easier to do a full reinstall of linux from e.g. FC4 or
Centos 4 repos than it is to get GCC 4* to build on an antique linux.
And don't forget the GSL and all the other core libraries, as well...

    rgb

>
> cheers!
> Chris
>

-- 
Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu