[gmx-users] Many failed tests in gromacs 4.6.3 installation

Wed Sep 11 14:23:17 CEST 2013

Various things changed after 4.6.1, such that the 4.6.1 tests cannot
be used to verify correctness of more recent versions. Unfortunately,
the install guide got out of sync there. I've changed the link so that
it directs you to the 4.6.3 test set. Please try that, and I expect
you'll have no problems.

It would be interesting to hear if you can get better mdrun
performance with icc 14 than you can with latest gcc.

Mark

On Wed, Sep 11, 2013 at 1:51 PM, Herbert Georg <hcgeorg at ufg.br> wrote:
> Hi there,
>
> I have installed a couple of days ago the latest (4.6.3) version of Gromacs,
> in a 12-core Xeon machine using the latest version (14.0) of the
> non-comercial Intel C++ compiler for linux.
>
> The installation seems to have succeeded, but I ran the latest set of tests
> available (4.6.1) and most of the tests PASSED, but still many of them
> FAILED.
>
> I've checked the tests that failed, and all of them are related to the
> grompp. Performing a <grep "FAILED"> in the output of the <./gmxtest.pl all>
> command gives the following:
>
> --------------------------------
> hcgeorg at araguaia:~/source/regressiontests-4.6.1> grep "FAILED" tests.log
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwBhamSh_GeomP1P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwBhamSh_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwBhamSh_GeomW3W3
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwBhamSh_GeomW4W4
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwLJSh_GeomP1P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwLJSh_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwLJSh_GeomW3W3
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwLJSh_GeomW4W4
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwNone_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSh_VdwNone_GeomW4P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwBhamSw_GeomP1P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwBhamSw_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwBhamSw_GeomW3W3
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwBhamSw_GeomW4W4
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwLJSw_GeomP1P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwLJSw_GeomW3W3
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwLJSw_GeomW4W4
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwNone_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecEwSw_VdwNone_GeomW4P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSh_GeomP1P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSh_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSh_GeomW3W3
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSh_GeomW4P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSh_GeomW4W4
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSw_GeomP1P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSw_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSw_GeomW3W3
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSw_GeomW4P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwBhamSw_GeomW4W4
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwLJSh_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwLJSh_GeomW3W3
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwLJSh_GeomW4P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwLJSh_GeomW4W4
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwLJSw_GeomW3P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwLJSw_GeomW3W3
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwLJSw_GeomW4P1
> FAILED. Check grompp.out files in nb_kernel_ElecRFCut_VdwLJSw_GeomW4W4
> 38 out of 142 kernel tests FAILED
> FAILED. Check grompp.out files in coulandvdwtogether
> FAILED. Check grompp.out files in expanded
> 2 out of 9 freeenergy tests FAILED
> --------------------------------------------------------
>
> Now, looking at the grompp.out files (or in the grompp.warn file), they all
> seem to have more then the 2 warnings that are in the reference.warn file.
> For instance:
>
> ---------------------------------------------------------
> hcgeorg at araguaia:~/source/regressiontests-4.6.1> cat
> kernel/nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1/grompp.warn
> WARNING 1 [file topol.top, line 103]:
>   The bond in molecule-type Protein between atoms 1 N and 2 H1 has an
>   estimated oscillational period of 1.0e-02 ps, which is less than 5 times
>   the time step of 2.0e-03 ps.
>   Maybe you forgot to change the constraints mdp option.
>
> WARNING 2 [file grompp.mdp]:
>   Can not exclude the lattice Coulomb energy between energy groups
>
> WARNING 3 [file grompp.mdp]:
>   The sum of the two largest charge group radii (0.077669) is larger than
>   rlist (0.900000) - rvdw (0.900000).
>   With exact cut-offs, better performance can be obtained with
>   cutoff-scheme = Verlet, because it does not use charge groups at all.
>
> WARNING 4 [file grompp.mdp]:
>   The sum of the two largest charge group radii (0.168825) is larger than
>   rlist (0.900000) - rcoulomb (0.900000).
>   With exact cut-offs, better performance can be obtained with
>   cutoff-scheme = Verlet, because it does not use charge groups at all.
>
> hcgeorg at araguaia:~/source/regressiontests-4.6.1> cat
> kernel/nb_kernel_ElecEwSw_VdwLJSw_GeomW3P1/reference.warn
> WARNING 1 [file topol.top, line 103]:
>   The bond in molecule-type Protein between atoms 1 N and 2 H1 has an
>   estimated oscillational period of 1.0e-02 ps, which is less than 5 times
>   the time step of 2.0e-03 ps.
>   Maybe you forgot to change the constraints mdp option.
>
> WARNING 2 [file grompp.mdp]:
>   Can not exclude the lattice Coulomb energy between energy groups
>
> ----------------------------------------------------------------
>
> Is this something to worry about? Or just a problem of difference versions
> giving different levels of warnings?
>
> Thanks in advance
>
> --
> Prof. Herbert de Castro Georg
> Instituto de Física - Universidade Federal de Goiás
> Campus Samambaia, CP 131, 74001-970, Goiânia, GO, Brasil
> Tel: +55 62 3521-1014 Ramal 209
> Fax: +55 62 3521-1014 Ramal 247
>
> --
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the www
> interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists