You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
When running on the parallelworks cluster, I seem to get invalid pe count errors for certain tests which go away when I change the pe count to the suggested count. Since these tests run fine elsewhere, I think the define domains routine might be missing the check on other systems.
To Reproduce
run on the parallelworks cluster, test_fms2_io.sh tests 4,5, and 7. also saw with test_global_arrays.2
Expected behavior mpp_define_domains2d should behave similarly across platforms and not error out
System Environment
Describe the system environment, include:
(parallelworks cluster, devcimultiintel)
OS: centos 7
Compiler/MPI: intel+impi/2022.1.2
netCDF Version: c 4.7.0, fortran 4.4.5
Configure options: none
The text was updated successfully, but these errors were encountered:
This was with gnu 13.2.0 instead of intel. It may be related to the testing harness script instead of some type of code issue, since that determines if the oversubscribe flag is passed to mpirun.
Describe the bug
When running on the parallelworks cluster, I seem to get invalid pe count errors for certain tests which go away when I change the pe count to the suggested count. Since these tests run fine elsewhere, I think the define domains routine might be missing the check on other systems.
To Reproduce
run on the parallelworks cluster, test_fms2_io.sh tests 4,5, and 7. also saw with test_global_arrays.2
Expected behavior
mpp_define_domains2d
should behave similarly across platforms and not error outSystem Environment
Describe the system environment, include:
(parallelworks cluster, devcimultiintel)
The text was updated successfully, but these errors were encountered: