Howard Pritchard
Howard Pritchard
we do need a fix for this on at least 4.1.x branch.
Problem for removing srun based launch of prrted's on systems of some interest like ORNL frontier and NERSC perlmutter is that we will lose support for using the OFI libfabric...
could use ```sinfo``` to get version info. not sure that's really any different though.
related to https://github.com/openpmix/prrte/pull/1972
This PR is just adding some protection. Someone else when they have time can go through the entire ompi code base and add similar protection. If this passes CI i''m...
related to #12336
I think the OFI problem in #8305 is misleading you. Can you post the output of ```ompi_info``` from both the laptop and cloud VM install?
Sorry for the delay in responding. I see you actually do have both ucx and ofi libfabric installed on the systems. To make sure we aren't trying to debug one...
Thanks. Now it would be useful to see if ucx transport works for both systems. Could you try first ``` stuff preceding mpirun --verbose -n 8 --mca pml ucx /slowdata/richardb/easybuild/build/OpenMPI/4.1.4/GCC-12.2.0/mpi_test_ring_c...
Thanks! This is pointing to an issue using OFI libfabric on your laptop. I'm not sure this is worth pursing further as we generally do not recommend using OFI libfabric...