Scheduled Downtime
On Tuesday 24 October 2023 @ 5pm MT the forums will be in read only mode in preparation for the downtime. On Wednesday 25 October 2023 @ 5am MT, this website will be down for maintenance and expected to return online later in the morning.
Normal Operations
The forums are back online with normal operations. If you notice any issues or errors related to the forums, please reach out to help@ucar.edu

Error with MPI

These days, I tried to build parallel CAM version 4.0 on a linux cluster, with gfortran-4.1.2, pgi-7.1.4, netcdf-3.6.2, mpich-1.2.7p1, etc, and a strange thing turned up:

When i use the command:
> $cfgdir/configure -dyn fv -hgrid 10x15 -nospmd -nosmp -test
it can find the netcdf library;

But, When i use the command:
> $cfgdir/configure -dyn fv -hgrid 1.9x2.5 -ntasks 6 -nosmp -test
it cannot find the mpi library, I have set the environment variables:
setenv INC_NETCDF /state/partition1/soft/libs/netcdf/3.6.2/gcc.pgf90/include
setenv LIB_NETCDF /state/partition1/soft/libs/netcdf/3.6.2/gcc.pgf90/lib
setenv MOD_NETCDF /state/partition1/soft/libs/netcdf/3.6.2/gcc.pgf90/include
setenv INC_MPI /data/dawning/mpich-1.2.7p1/include
setenv LIB_MPI /data/dawning/mpich-1.2.7p1/lib

The error message was:
Issuing command to the CICE configure utility:
/lustre/users/essljn/CSM4/ccsm4_0_01/models/ice/cice/bld/configure -hgrid 1.9x2.5 -cice_mode prescribed -ntr_aero 0 -ntr_pond 1 -ntr_iage 0 -ntasks 6 -nthreads 1 -cache config_cache_cice.xml -cachedir /lustre/users/essljn/CSM4/ccsm4_0_01/fv1.9x2.5_01
configure done.
creating /lustre/users/essljn/CSM4/ccsm4_0_01/fv1.9x2.5_01/Filepath
creating /lustre/users/essljn/CSM4/ccsm4_0_01/fv1.9x2.5_01/misc.h
creating /lustre/users/essljn/CSM4/ccsm4_0_01/fv1.9x2.5_01/preproc.h
creating /lustre/users/essljn/CSM4/ccsm4_0_01/fv1.9x2.5_01/Makefile
creating /lustre/users/essljn/CSM4/ccsm4_0_01/fv1.9x2.5_01/config_cache.xml
Looking for a valid GNU make... using gmake
Test linking to NetCDF library... ok
Test linking to MPI library... **** FAILED ****
Issued the command:
gmake -f /lustre/users/essljn/CSM4/ccsm4_0_01/fv1.9x2.5_01/Makefile test_mpi 2>&1
The output was:
......

Any help will be appreciate!

Best regards.

Leo
 

eaton

CSEG and Liaisons
Was the mpi lib built using gfortran? My guess is that that is possibly the problem. Typically you want to build the netcdf and mpi libs using the same fortran compiler that is used to build cam. The output from the failed test which you omitted in your post may also contain clues.
 
Top