Dear CFDEM developers:
I am now trying to run the latest CFDEM on an HPC cluster however after being able to compile LIGGGHTS, OpenFOAM and CFDEM the ErgunTestMPI fails with the below error:
...
[1] #7 ???? in "/project/fchen14/CFDEM/CFDEMcoupling-PUBLIC-5.x/platforms/linux64IccDPInt32Opt/bin/cfdemSolverPiso"
[smic037:mpi_rank_0][error_sighandler] Caught error: Floating point exception (signal 8)
I am on a RHEL 6.9 cluster with
icc version 18.0.0
LIGGGHTS version 3.8.0
Latest OF5.x
CFDEMcoupling version: cfdem-3.8.1
, compatible to LIGGGHTS version: 3.8.0
, compatible to OF version and build: 5.x-commit-538044ac05c4672b37c7df607dca1116fa88df88
MPI version:
I tried intel MPI, impi/2018.0.128/intel64 and
also MVAPICH2.3, compiled using icc version 18.0.0
Because it is a cluster environment so everything is compiled from source, I have tried the combination of IntelMPI/MVAPICH2 + OF5.0/OF5.x, all resulted in the same floating point exception so I am wondering if there is any suggestions to my environment or I need to modify the example tutorial files as well?
I have attached my ErgunTestMPI case folder with all the error logs.
Thanks for any suggestions!
Roy
| Attachment | Size |
|---|---|
 case file with logs | 3.69 MB |
paul | Fri, 08/17/2018 - 16:36
The relevant line is here:
The relevant line is here:
[2] #3 Foam::twoWayMPI::setCG() const[1] #3 Foam::twoWayMPI::setCG() const[3] #3 Foam::twoWayMPI::setCG() const in "/project/fchen14/CFDEM/CFDEMcoupling-PUBLIC-5.x/platforms/linux64IccDPInt32Opt/lib/liblagrangianCFDEM-PUBLIC-5.x.so"
Did you attempt coarse graining?
dbxmcf | Fri, 08/17/2018 - 18:20
Hi, Paul,
Hi, Paul,
Thanks for the response, I did not change anything but just cloned the CFDEM repository from github, I am not aware of that coarse graining option at this moment, if it is the cause, could you tell where to change it?
Thanks,
Roy