[gmx-users] Compiling GROMACS with FORTRAN?

[Mike Hanby]

Howdy Ivano,

Would you mind sharing the steps used to compile Gromacs on the
BlueGene/L?

I've tried this several times with the IBM cross compilers and always
end up failing.

Thanks for any information,

Mike

-----Original Message-----
From: gmx-users-bounces at gromacs.org
[mailto:gmx-users-bounces at gromacs.org] On Behalf Of Ivano Eberini
Sent: Friday, October 12, 2007 05:31
To: Discussion list for GROMACS users
Subject: Re: [gmx-users] Compiling GROMACS with FORTRAN?

Il giorno 12/ott/07, alle ore 11:29, David van der Spoel ha scritto:

> Ivano Eberini wrote:
>> Dear all,
>> we have just installed GROMACS-3.3.2 on a BlueGene (BG/L) cluster,  
>> compiling it without FORTRAN.
>> Do you deem it could be better to recompile it adding "--enable- 
>> fortran"?
> the proof of the pudding is in the eating.
> please try it.
>

Dear David,
here are the results for two runs (full.mdp in speptide folder of  
GROMACS tutorial) with 16 chips / 32 tasks on BG/L,

without FORTRAN:


                NODE (s)   Real (s)      (%)
        Time:    161.000    161.000    100.0
                        2:41
                (Mnbf/s)   (GFlops)   (ns/day)  (hour/ns)
Performance:     12.529      3.034      5.366      4.472


with FORTRAN:

                NODE (s)   Real (s)      (%)
        Time:    155.000    155.000    100.0
                        2:35
                (Mnbf/s)   (GFlops)   (ns/day)  (hour/ns)
Performance:     13.021      3.143      5.574      4.306

Best regards,
Ivano
--
Ivano Eberini
Gruppo di Studio per la Proteomica e la Struttura delle Proteine
Dipartimento di Scienze Farmacologiche
Università degli Studi di Milano
Via Giuseppe Balzaretti, 9
20133 - Milano
tel.: +39-02-50318395, fax: +39-02-50318284

"Quelli che scavano il sole, cercando un'ombra migliore"