[CIG-SEISMO] Compilation error for specfem3D_globe / PS
Dimitri Komatitsch
komatitsch at lma.cnrs-mrs.fr
Thu Jun 2 15:51:43 PDT 2016
Hi,
That strongly depends on the max frequency you want (scales as the
fourth power of that).
Cheers,
Dimitri.
On 02/06/2016 08:52, Julien Chaput wrote:
> Out of curiosity, how realistic would it be to produce 60000+ seconds of
> seismogram on 24 cores? From what I can see from early tests, it could
> take upwards of 6 months (naturally unfeasible), unless there are tricks
> to speed things up?
> Cheers
>
> Julien
>
> On Wed, Jun 1, 2016 at 11:28 PM, Julien Chaput <jchaput82 at gmail.com
> <mailto:jchaput82 at gmail.com>> wrote:
>
> Good question; I jumped straight to 5.3.0 from 4.7.7, so perhaps
> 5.3.0 might not be the minimum. Though, supposedly 4.7.7 also had
> fortran 2008 compatibility (from the info I could find, anyways).
>
> On Wed, Jun 1, 2016 at 5:49 PM, Dimitri Komatitsch
> <komatitsch at lma.cnrs-mrs.fr <mailto:komatitsch at lma.cnrs-mrs.fr>> wrote:
>
>
>
> Thanks!
>
> What about gcc/gfortran 4.8?
> That is what many users have (most Linux distributions have 4.8
> or 4.9).
>
> Thanks,
> Dimitri.
>
> On 02/06/2016 01:47, Julien Chaput wrote:
>
> Okay, it looks like everything compiles on gcc 5.3.0 at a
> minimum. It
> might perhaps be useful to add a note in the manual?
> Cheers
>
> Julien
>
> On Tue, May 31, 2016 at 1:05 AM, Daniel B. Peter
> <daniel.peter at kaust.edu.sa
> <mailto:daniel.peter at kaust.edu.sa>
> <mailto:daniel.peter at kaust.edu.sa
> <mailto:daniel.peter at kaust.edu.sa>>> wrote:
>
> Hi all,
>
> Dimitri is right, that’s 2008 standard. the contiguous
> keyword is
> implemented by most compilers by now, so please try
> updating the
> compiler.
>
> when choosing —enable-debug, it tests for standard 2008
> or the gnu
> extension in case gfortran is used, that’s why buildbot
> is allowing it.
>
> best wishes,
> daniel
>
>
> On May 31, 2016, at 3:38 AM, Julien Chaput
> <jchaput82 at gmail.com <mailto:jchaput82 at gmail.com>
> <mailto:jchaput82 at gmail.com
> <mailto:jchaput82 at gmail.com>>> wrote:
>
> Hi,
>
> Thanks for the reply. I'm currently running gcc
> 4.4.7 on Centos 6.
> I suspect it's probably time to update to the
> latest standard,
> though I'm glad it should be an easy fix. If the
> error persists
> after the upgrade, I'll let you know.
> Cheers
>
> Julien
>
> On Mon, May 30, 2016 at 5:49 PM, Dimitri Komatitsch
> <komatitsch at lma.cnrs-mrs.fr
> <mailto:komatitsch at lma.cnrs-mrs.fr>
> <mailto:komatitsch at lma.cnrs-mrs.fr
> <mailto:komatitsch at lma.cnrs-mrs.fr>>>
>
> wrote:
>
>
> Hi again all,
>
> PS for David: BuildBot could/should have found
> this I guess
> because we compile with -f2003 when we check
> the standard (at
> least with Intel ifort, also with GNU gfortran
> I think), thus
> we should have seen an error for that new
> keyword that is in
> f2008 but not in f2003.
>
> Weird...
>
> thanks
> Dimitri
>
> On 31/05/2016 01:46, Dimitri Komatitsch wrote:
>
>
> Hi Julien, Hi all,
>
> Probably your Fortran compiler that does
> not support
> Fortran2008, and
> Daniel (cc'ed) recently added the
> "contiguous" keyword,
> which is
> specific to F2008.
>
> Daniel, could you maybe remove it? We need
> to make sure
> the code remains
> portable.
>
> Thanks,
> Dimitri.
>
> On 31/05/2016 00:22, Julien Chaput wrote:
>
> Hi,
>
> I'm a newbie with specfem3D_globe, and
> am attempting
> to generate long
> coda (80000s) for arbitrary scattering
> structures to
> study the character
> of coda correlation functions.
> I'm currently running into a
> compilation problem (i.e.
> make specfem3D),
> though the mesher compiles without issues.
> The error I can't seem to get around is
> this:
>
> src/specfem3D/sepcfem3D_par.F90:402.25
> real(kind=CUSTOM_REAL), dimension(:),
> pointer,
> contiguous ::
> b_rmassz_crust_m
> 1
> Error: Invalid character in name at (1)
>
> This error is flagged at lines 402,
> 403, 404, 434, 438
> 581, 582, 612,
> 616 (before the error count maxes out).
>
> I'm currently using openmpi running on
> 24 cores, and
> have configured
> simply as:
> ./configure FC=mpif90 MPIFC=mpif90
> (I've also tried setting FC=gfortran to
> no avail).
> Has anyone encountered this sort of
> thing? As best I
> can understand (I'm
> not a native fortran programmer) it
> seems to be
> related to a syntax
> clash between fortran compilers.
> Thanks
>
> Julien Chaput
> Postdoctoral fellow
> Colorado State University
> jchaput82 at gmail.com <mailto:jchaput82 at gmail.com>
> <mailto:jchaput82 at gmail.com <mailto:jchaput82 at gmail.com>>
> <mailto:jchaput82 at gmail.com
> <mailto:jchaput82 at gmail.com> <mailto:jchaput82 at gmail.com
> <mailto:jchaput82 at gmail.com>>>
>
>
>
>
>
>
> _______________________________________________
> CIG-SEISMO mailing list
> CIG-SEISMO at geodynamics.org
> <mailto:CIG-SEISMO at geodynamics.org>
> <mailto:CIG-SEISMO at geodynamics.org
> <mailto:CIG-SEISMO at geodynamics.org>>
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-seismo
>
>
>
> --
> Dimitri Komatitsch
> CNRS Research Director (DR CNRS), Laboratory of
> Mechanics and
> Acoustics,
> UPR 7051, Marseille, France
> http://komatitsch.free.fr
> <http://komatitsch.free.fr/>
>
>
>
>
>
> ------------------------------------------------------------------------
>
> This message and its contents including attachments are
> intended
> solely for the original recipient. If you are not the
> intended
> recipient or have received this message in error,
> please notify me
> immediately and delete this message from your computer
> system. Any
> unauthorized use or distribution is prohibited. Please
> consider the
> environment before printing this email.
>
>
>
> --
> Dimitri Komatitsch
> CNRS Research Director (DR CNRS), Laboratory of Mechanics and
> Acoustics,
> UPR 7051, Marseille, France http://komatitsch.free.fr
>
>
>
--
Dimitri Komatitsch
CNRS Research Director (DR CNRS), Laboratory of Mechanics and Acoustics,
UPR 7051, Marseille, France http://komatitsch.free.fr
More information about the CIG-SEISMO
mailing list