Thank you. You were right. Switching to another openmpi version (2.2.1)
solved the problem: pmemd.MPI was built and pmemd tests passed.
(Problems were with openmpi 1.4.3)
[kmirosh.gdeveloper ~]$ mpif90 -show
gfortran -I/opt/openmpi/2.1.1/include -pthread -I/opt/openmpi/2.1.1/lib
-Wl,-rpath -Wl,/opt/openmpi/2.1.1/lib -Wl,--enable-new-dtags
-L/opt/openmpi/2.1.1/lib -lmpi_usempi -lmpi_mpifh -lmpi
AmberTools were successfully compiled with -mpi also but some tests gave
errors:
cd 07_Comprehensive && ./Run.comprehensive
./Run.comprehensive: Program error
make[3]: [COMPREHENSIVE] Error 1 (ignored)
cd PIMD_Langevin_LFMiddle_water && ./Run.LGV_LF
AMBERHOME is /home/kmirosh/Programs/amber18
Testing PIMD with langevin dynamics
[gnode9:08311] 1 more process has sent help message help-mpi-api.txt /
mpi-abort
[gnode9:08311] Set MCA parameter "orte_base_help_aggregate" to 0 to see all
help / error messages
./Run.LGV_LF: Program error
make[2]: [test.sander.BASIC.MPI] Error 1 (ignored)
cd PIMD_Constr_Andersen_LFMiddle_water && ./Run.ADS_LF
AMBERHOME is /home/kmirosh/Programs/amber18
Testing PIMD with Andersen thermostat
[gnode9:08328] 1 more process has sent help message help-mpi-api.txt /
mpi-abort
[gnode9:08328] Set MCA parameter "orte_base_help_aggregate" to 0 to see all
help / error messages
./Run.ADS_LF: Program error
make[2]: [test.sander.BASIC.MPI] Error 1 (ignored)
regards,
Kateryna
сб, 19 січ. 2019 о 16:26 David Case <david.case.rutgers.edu> пише:
> On Fri, Jan 18, 2019, Kateryna Miroshnychenko wrote:
>
> >I've a problem with parallel (MPI) installation of pmemd18 on cluster. The
> >serial install went smoothly.
> >May by the problem is in the old gcc(4.4.6)?
>
> It seems more likely that something is amiss with your mpi installation.
>
> 1. Do other parallel programs (like sander.MPI) build and run OK?
> 2. How did you install or load MPI? Which version is it?
> 2. what is the result of typing "mpif90 -show"?
>
> >Error: Symbol 'mpi_double' at (1) has no IMPLICIT type
>
> The "mpi_double" variable comes from the MPI installation, via the
> mpif90 wrapper. Hence the questions above.
>
> ...good luck....dac
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Jan 21 2019 - 14:00:02 PST
