Hello,I am running a model in EE10.3, and I have a problem with MPI. If I run the model with a single domain, it works. But if I try to run it with several subdomains, the model stops with the following error:"Abort(608247566) on node 1 (rank 1 in comm 0): Fatal error in PMPI_Bcast: Message truncated, error stack:PMPI_Bcast(431)…: MPI_Bcast(buf=000000CF7B975A00, count=1, dtype=0x4c000427, root=0, comm=comm=0x84000002) failedPMPI_Bcast(417)…:MPIDI_Bcast_intra_composition_gamma(397):MPIDI_NM_mpi_bcast(149)…:MPIR_Bcast_intra_auto(222)…:MPIR_Bcast_intra_binomial(111)…:(unknown)(): Message truncated"I tried the following different cases:- the whole model is decomposed into 2 subdomains in j direction (P0, P1). Subdomain P1 is not active, because its southern neighbor is 0. - the whole model is decomposed into 2 subdomains in i direction (P0, P1). Subdomain P1 is not active, because its western neighbor is 0.Could anyone please help me figure out how to solve this issue?Thanks,Mohammad
Which version of EEMS are you using?
Can you provide your e-mail to me in a private message so I can better support you?
Another preliminary question is, are you using EE 10.3 to create the decomp.jnp file?
Hi Tom, Yes I am using EE10.3, and I create decomp.jnp using EE10.3. My Email is “email@example.com” please feel free to email me. Thanks,
If possible, please send your model input files to firstname.lastname@example.org.
We will take a look at the model setup and get back to you regarding this error.
We have determined that there is an issue when using MPI with spatially varying AHD. We have fixed this and will provide the fix in the next update which will be 10.3.2 in the next week or so.