search

SPECFEM / specfem2d

SPECFEM2D simulates forward and adjoint seismic wave propagation in two-dimensional acoustic, (an)elastic, poroelastic or coupled acoustic-(an)elastic-poroelastic media, with Convolution PML absorbing conditions.
GNU General Public License v3.0
204 stars 147 forks source link

specfem2d nightly build has failed #755

Closed buildbot-princeton closed 7 years ago

buildbot-princeton commented 7 years ago

See build 272 and PDFs.

luet commented 7 years ago

I posted the first files that differ:

The problem start at L72: -0.7740000E-01 -0.2696017-232 where it's 0 in the reference seismograms. My guess would be that a variable was not initialized to 0.

komatits commented 7 years ago

Hi David,

Thanks! I think it's Alexis' last commit that broke that, probably easy to fix, Alexis will fix it tomorrow (today is a national holiday here in France, not because we celebrate our new President but because of the end of WWII ;-)

Thanks, Best,

Dimitri.

On 05/08/2017 05:39 PM, David Luet wrote:

I posted the first files that differ:

The problem start at L72: |-0.7740000E-01 -0.2696017-232| where it's 0 in the reference seismograms. My guess would be that a variable was not initialized to 0.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/geodynamics/specfem2d/issues/755#issuecomment-299904058, or mute the thread https://github.com/notifications/unsubscribe-auth/AFjDKUAjQQRjHCgndtyjg747BxrBrxsQks5r3zczgaJpZM4NTZfN.

-- Dimitri Komatitsch, CNRS Research Director (DR CNRS) Laboratory of Mechanics and Acoustics, Marseille, France http://komatitsch.free.fr

komatits commented 7 years ago

Hello all, I doubt that my commit have broken the code, it consisted in removing some useless comments! I have also updated the reference seismograms in the EXAMPLES/axisymmetric_case_AXISYM_option.

David in the file you posted http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_latest some numbers are not written correctly (-0.2696017-232 does not make sense). I am not able to reproduce that file. I just made a git pull and everything works on my side in debug/release/MPI/serial (no strange numbers in AA.S0002.BXX.semd and the fit with the reference is perfect)

I don't know where the problem come from...

Cheers,

Alexis Bottero, PhD student Tel: +0033 6 95 17 00 97 Waves and imaging Laboratoire de Mécanique et d'Acoustique de Marseille

2017-05-08 18:12 GMT+02:00 Dimitri Komatitsch komatitsch@lma.cnrs-mrs.fr:

Hi David,

Thanks! I think it's Alexis' last commit that broke that, probably easy to fix, Alexis will fix it tomorrow (today is a national holiday here in France, not because we celebrate our new President but because of the end of WWII ;-)

Thanks, Best,

Dimitri.

On 05/08/2017 05:39 PM, David Luet wrote:

I posted the first files that differ:

The problem start at L72: |-0.7740000E-01 -0.2696017-232| where it's 0 in the reference seismograms. My guess would be that a variable was not initialized to 0.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/geodynamics/specfem2d/issues/755# issuecomment-299904058, or mute the thread https://github.com/notifications/unsubscribe-auth/AFjDKUAjQ QRjHCgndtyjg747BxrBrxsQks5r3zczgaJpZM4NTZfN.

-- Dimitri Komatitsch, CNRS Research Director (DR CNRS) Laboratory of Mechanics and Acoustics, Marseille, France http://komatitsch.free.fr

komatits commented 7 years ago

Hello all,

I think I know, a few days ago I noticed someone had changed a unformattted ASCII write to a formatted one, that could be the problem. I will revert that (probably on Friday or next week, I am traveling until then)

cheers dimitri

On 05/09/2017 12:38 PM, Alexis Bottero wrote:

Hello all, I doubt that my commit have broken the code, it consisted in removing some useless comments! I have also updated the reference seismograms in the EXAMPLES/axisymmetric_case_AXISYM_option.

David in the file you posted http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_latest some numbers are not written correctly (|-0.2696017-232| does not make sense). I am not able to reproduce that file. I just made a git pull and everything works on my side in debug/release/MPI/serial (no strange numbers in AA.S0002.BXX.semd and the fit with the reference is perfect)

I don't know where the problem come from...

Cheers,

Alexis Bottero, PhD student Tel: +0033 6 95 17 00 97 Waves and imaging Laboratoire de Mécanique et d'Acoustique de Marseille

2017-05-08 18:12 GMT+02:00 Dimitri Komatitsch <komatitsch@lma.cnrs-mrs.fr mailto:komatitsch@lma.cnrs-mrs.fr>:

Hi David,

Thanks! I think it's Alexis' last commit that broke that, probably
easy to fix, Alexis will fix it tomorrow (today is a national
holiday here in France, not because we celebrate our new President
but because of the end of WWII ;-)

Thanks,
Best,

Dimitri.

On 05/08/2017 05:39 PM, David Luet wrote:

    I posted the first files that differ:

      * AA.S0002.BXX.semd reference

    <http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_reference
    <http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_reference>>
      * AA.S0002.BXX.semd latest

    <http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_latest
    <http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_latest>>

    The problem start at L72: |-0.7740000E-01 -0.2696017-232
    <tel:2696017-232>| where it's 0
    in the reference seismograms. My guess would be that a variable
    was not
    initialized to 0.

    —
    You are receiving this because you are subscribed to this thread.
    Reply to this email directly, view it on GitHub
    <https://github.com/geodynamics/specfem2d/issues/755#issuecomment-299904058
    <https://github.com/geodynamics/specfem2d/issues/755#issuecomment-299904058>>,
    or mute the thread
    <https://github.com/notifications/unsubscribe-auth/AFjDKUAjQQRjHCgndtyjg747BxrBrxsQks5r3zczgaJpZM4NTZfN
    <https://github.com/notifications/unsubscribe-auth/AFjDKUAjQQRjHCgndtyjg747BxrBrxsQks5r3zczgaJpZM4NTZfN>>.

--
Dimitri Komatitsch, CNRS Research Director (DR CNRS)
Laboratory of Mechanics and Acoustics, Marseille, France
http://komatitsch.free.fr

-- Dimitri Komatitsch, CNRS Research Director (DR CNRS) Laboratory of Mechanics and Acoustics, Marseille, France http://komatitsch.free.fr

komatits commented 7 years ago

Hello all,

I think I know why, a few days ago I noticed someone had changed a unformattted ASCII write to a formatted one, that could be the problem. I will revert that (probably on Friday or next week, I am traveling until then)

cheers dimitri

On 05/09/2017 12:38 PM, Alexis Bottero wrote:

Hello all, I doubt that my commit have broken the code, it consisted in removing some useless comments! I have also updated the reference seismograms in the EXAMPLES/axisymmetric_case_AXISYM_option.

David in the file you posted http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_latest some numbers are not written correctly (|-0.2696017-232| does not make sense). I am not able to reproduce that file. I just made a git pull and everything works on my side in debug/release/MPI/serial (no strange numbers in AA.S0002.BXX.semd and the fit with the reference is perfect)

I don't know where the problem come from...

Cheers,

Alexis Bottero, PhD student Tel: +0033 6 95 17 00 97 Waves and imaging Laboratoire de Mécanique et d'Acoustique de Marseille

2017-05-08 18:12 GMT+02:00 Dimitri Komatitsch <komatitsch@lma.cnrs-mrs.fr mailto:komatitsch@lma.cnrs-mrs.fr>:

Hi David,

Thanks! I think it's Alexis' last commit that broke that, probably
easy to fix, Alexis will fix it tomorrow (today is a national
holiday here in France, not because we celebrate our new President
but because of the end of WWII ;-)

Thanks,
Best,

Dimitri.

On 05/08/2017 05:39 PM, David Luet wrote:

    I posted the first files that differ:

      * AA.S0002.BXX.semd reference

    <http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_reference
    <http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_reference>>
      * AA.S0002.BXX.semd latest

    <http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_latest
    <http://geoweb.princeton.edu/%7Eluet/specfem2d_nightly_fail/AA.S0002.BXX.semd_latest>>

    The problem start at L72: |-0.7740000E-01 -0.2696017-232
    <tel:2696017-232>| where it's 0
    in the reference seismograms. My guess would be that a variable
    was not
    initialized to 0.

    —
    You are receiving this because you are subscribed to this thread.
    Reply to this email directly, view it on GitHub
    <https://github.com/geodynamics/specfem2d/issues/755#issuecomment-299904058
    <https://github.com/geodynamics/specfem2d/issues/755#issuecomment-299904058>>,
    or mute the thread
    <https://github.com/notifications/unsubscribe-auth/AFjDKUAjQQRjHCgndtyjg747BxrBrxsQks5r3zczgaJpZM4NTZfN
    <https://github.com/notifications/unsubscribe-auth/AFjDKUAjQQRjHCgndtyjg747BxrBrxsQks5r3zczgaJpZM4NTZfN>>.

--
Dimitri Komatitsch, CNRS Research Director (DR CNRS)
Laboratory of Mechanics and Acoustics, Marseille, France
http://komatitsch.free.fr

-- Dimitri Komatitsch, CNRS Research Director (DR CNRS) Laboratory of Mechanics and Acoustics, Marseille, France http://komatitsch.free.fr

luet commented 7 years ago

I use the following compile options:

./configure FC=ifort CC=icc MPIFC=mpif90 MPICC=mpicc --with-mpi --enable-double-precision

I went back in history and it seems that it's commit 406b53c that broke it, 406b53c runs fine.

komatits commented 7 years ago

Hi David,

thanks! Good point, it is this I was referring to as well but not remembering in which routine it was:

changed to

We should just write: write(11,*)

I will commit that tonight from the hotel of the conference.

thanks cheers dimitri

On 05/09/2017 08:01 PM, David Luet wrote:

I use the following compile options:

|./configure FC=ifort CC=icc MPIFC=mpif90 MPICC=mpicc --with-mpi --enable-double-precision |

I went back in history and it seems that it's commit 406b53c https://github.com/geodynamics/specfem2d/commit/406b53c3eb75d5f21bf4e24e46338b2034b1946d that broke it, 406b53c https://github.com/geodynamics/specfem2d/commit/406b53c3eb75d5f21bf4e24e46338b2034b1946d runs fine.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/geodynamics/specfem2d/issues/755#issuecomment-300251398, or mute the thread https://github.com/notifications/unsubscribe-auth/AFjDKVduYpJRh15NpFDyAiqCREYw0d3Iks5r4KoOgaJpZM4NTZfN.

-- Dimitri Komatitsch, CNRS Research Director (DR CNRS) Laboratory of Mechanics and Acoustics, Marseille, France http://komatitsch.free.fr

luet commented 7 years ago

The problem is fixed.