September 2019 - discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] test/mpi/dtpools/configure bashism
by Zhou, Hui 17 Sep '19

17 Sep '19

Thanks for the patch. FYI, The code in the master has already refactored the mentioned lines away. — Hui Zhou On Sep 16, 2019, at 8:41 PM, Thomas Jahns via discuss <discuss(a)mpich.org<mailto:[email protected]>> wrote: Hello, unfortunately mpich 3.3.1 configure does not run in BSD env: ./configure: types+=MPI_CHAR,: not found and so forth, please see attached build output log for more info. I've prepared a patch to test/mpi/dtpools/configure.ac to use portable sed syntax instead of the bash subshell syntax which is causing problems in this case. Regards, Thomas <configure-bashism.patch><spack-build-out.txt.gz>_______________________________________________ discuss mailing list discuss(a)mpich.org<mailto:[email protected]> To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

test/mpi/dtpools/configure bashism
by Thomas Jahns 16 Sep '19

16 Sep '19

Hello, unfortunately mpich 3.3.1 configure does not run in BSD env: ./configure: types+=MPI_CHAR,: not found and so forth, please see attached build output log for more info. I've prepared a patch to test/mpi/dtpools/configure.ac to use portable sed syntax instead of the bash subshell syntax which is causing problems in this case. Regards, Thomas

1 0

Heads up for argument mismatch error with gfortran 10
by Thomas König 16 Sep '19

16 Sep '19

Hi, this is just a heads-up: Version 10 of the gfortran compiler will no longer accept mismatched arguments to old-style Fortran routines, see https://gcc.gnu.org/gcc-10/changes.html and https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91556 Currently, this leads to a configure error with mpich 3.3.1: checking whether gfortran allows mismatched arguments... no configure: error: The Fortran compiler gfortran will not compile files that call the same routine with arguments of different types. There is now a -fallow-argument-mismatch option which degrades the error to a warning, which you may want to put into your configure script. Alternatively, gfortran also provides the ISO_Fortran_binding.h file since its version 9, you might also want to use that. The release of gfortran 10 will be quite some time in the future, so there is no immediate rush. Regards Thomas

1 0

Re: [mpich-discuss] Running on Cluster issue
by Raffenetti, Kenneth J. 06 Sep '19

06 Sep '19

This is often indicative of a mismatched MPI library and mpiexec launcher. Take care to ensure that the MPI library linked to your binary and mpiexec are (ideally) part of the same installation. Ken On 9/6/19 12:34 AM, Sepinoud Azimi via discuss wrote: > Hi, > > I am trying to run a python code on cluster with Mpiexec using the command mpiexec -n 5 python parallel_simulation.py. This is supposed to run on 5 processes (that is what happening on local machine). But on cluster it keeps running it with 1 process. I have tried mpi4py but apparently it has conflict. I was wondering if you have any idea how to fix this. > > Many thanks in advance. > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss >

1 0

Running on Cluster issue
by Sepinoud Azimi 06 Sep '19

06 Sep '19

Hi, I am trying to run a python code on cluster with Mpiexec using the command mpiexec -n 5 python parallel_simulation.py. This is supposed to run on 5 processes (that is what happening on local machine). But on cluster it keeps running it with 1 process. I have tried mpi4py but apparently it has conflict. I was wondering if you have any idea how to fix this. Many thanks in advance.

1 0