5
$\begingroup$

I am using an AMD Linux cloud-based instance to ran general optimisation and transition state calculations in parallel. I have noticed that I can run gas phase calculations in parallel but the run fails when I include a solvent in the input file. Additionally, I have a local Intel workstation, and I face no issues on there. The same input file that fails on my cloud instance, works on my local workstation.

Log file:

> An error occurred in MPI_Type_match_size reported by process
> [4120707073,1] on communicator MPI_COMM_WORLD 
> 
> MPI_ERR_ARG: invalid argument of some other kind MPI_ERRORS_ARE_FATAL
> (processes in this communicator will now abort, ***    and potentially
> your MPI job) 
> 
> PMIX ERROR: UNREACHABLE in file server/pmix_server.c at line 2198 6
> more processes have sent help message help-mpi-errors.txt /
> mpi_errors_are_fatal 
> 
> Set MCA parameter "orte_base_help_aggregate" to 0 to see all help /
> error messages[file orca_tools/qcmsg.cpp, line 465]:  .... aborting
> the run

Failure in output file:

--------------------
CPCM SOLVATION MODEL
--------------------
CPCM parameters:
  Epsilon                                         ...      80.4000
  Refrac                                          ...       1.3300
  Rsolv                                           ...       1.3000
  Surface type                                    ... GAUSSIAN VDW
  Epsilon function type                           ...         CPCM
Radii:
 Radius for C  used is    3.8550 Bohr (=   2.0400 Ang.)
 Radius for O  used is    3.4469 Bohr (=   1.8240 Ang.)
 Radius for H  used is    2.4944 Bohr (=   1.3200 Ang.)
Calculating surface                               ...        done! (  0.0s)
GEPOL surface points                              ...         1252
GEPOL Volume                                      ...    1244.2933
GEPOL Surface-area                                ...     725.8299
Calculating surface distance matrix               ...        done! (  0.0s)
Performing Cholesky decomposition & store         ...        
ORCA finished by error termination in SCF
Calling Command: mpirun -np 16  /opt/orca503/orca_scf_mpi g_acoh_gp.gbw b g_acoh_gp 
[file orca_tools/qcmsg.cpp, line 465]: 
  .... aborting the run



Orca version: Program Version 5.0.3 -  RELEASE  -

Open MPI version: mpiexec --version: mpiexec (OpenRTE) 4.1.1

AMD processor: AMD EPYC Processor (with IBPB)

Intel Processor: Intel(R) Xeon(R) CPU E5-2650 v4 @ 2.20GHz

I am not using a scheduler, running the Orca command at the prompt.

What could be the issue?

Improve this question
$\endgroup$
6
  • 1
    $\begingroup$ Orca runs on AMD processor on my laptop without any problem, so my guess is that it has something to do with the openmpi. Maybe the OpenMPI was compiled with different integer sizes or something else is different. Or maybe there is something wrong with file I/O because on the solvent based run I believe Orca writes some files with the tessellation information. $\endgroup$
    – S R Maiti
    May 13, 2022 at 13:31
  • 1
    $\begingroup$ Can you post a few tens of lines before "ORCA finished by error termination in SCF"? This way we can see exactly what happened right before the termination. $\endgroup$
    – wzkchem5
    May 13, 2022 at 20:11
  • $\begingroup$ @wzkchem5 Added a few preceding lines - it is failing after beginning the CPCM calculations (Cholesky decomposition and store in particular) $\endgroup$
    – learner254
    May 13, 2022 at 21:00
  • $\begingroup$ @SRMaiti If this is the case, would OpenMPI recompilation solve the issue? is there a flag/option I should use when recompiling? I'll give it a go. Also would you have a link where I can read more on Orca's solvation method under the hood? $\endgroup$
    – learner254
    May 13, 2022 at 21:10
  • 1
    $\begingroup$ Yes, you could try recompiling OpenMPI, that might work. There might also be some problem with BLAS libraries as well (because it stops at Cholesky decomposition). Honestly, it could be anything, difficult to say from the error message. You should post this issue on the orcaforum, the developers might be able to help. Orca is closed-source so I don't think there is any way of knowing how they implement solvation (apart from what's in the manual). $\endgroup$
    – S R Maiti
    May 14, 2022 at 22:41

1 Answer 1

Reset to default
3
$\begingroup$

Turns out the issue is that Fortran is required to run solvent-based calculations in parallel. My cloud instances did not have it installed. Problem was solved after installing Fortran. Answer was from Orca Forum

Improve this answer
$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .