[phyczz]

I build a model using OpenFOAM and try to run the model in parallel.
When the size of the mesh is about 700M, the program goes well.
When I improve the number of cells and the size of the mesh becomes 5G some errors happen.
I tried to use 8 cores for parallel calculation, the errors are:
(1)7 more processes have sent help message help-btl-vader.txt / cma-permission-denied
(2)mpirun detected that one or more processes exited with non-zero status, thus causing
the job to be terminated.

Could you give me some help about how to solve this problem?

p.s. The size of the mesh which generated by blockMesh is about 5G.
The solver I used is rhoPimpleFoam.
My computer has 8 cores and 16G RAM.

[klausb]

I suggest to run a "rhoPimpleFoam" tutorial case to check, whether there's a problem with your installation or permissions.

If possible use one of the tutorial cases as the basis for your case setup.

[phyczz]

Thank you klausb!

When the size of the mesh is about 700M, the program goes well.
When I improve the number of cells and the size of the mesh becomes 5G some errors happen.

Is there a limit of the mesh size during the calculation?

[klausb]

A lot of cells per core but apparently it's not an OpenFoam problem but an OpenMPI "issue" see: https://github.com/conda-forge/mpi4p...tock/issues/15


With a lot of cells per core, a lot of information needs to be exchanged between cores maybe that's triggering issues on your hardware. Follow the advise suggested in the blog linked above and/or consider an OpenFoam source installation using MPICH or another MPI implementation.

[phyczz]

Thank you so much for your reply.
I tried MPICH but the problem remains.
Maybe there is a limitation of the cell number for a specific computer.

In addtion, I met another problem.
When I used blockMesh to create mesh and the cell number was about 60,000,000
then the program crashed.

The error information are:
"#0 blockMesh: malloc.c:2401: sysmalloc: Assertion `(old_top == initial_top (av) && old_size == 0) || ((unsigned long) (old_size) >= MINSIZE && prev_inuse (old_top) && ((unsigned long) old_end & (pagesize - 1)) == 0)' failed.
Aborted (core dumped)"


Will other mesh generation utility be better when deal with mesh than has large cell number?

Thank you very much!

[hokhay]

When you said "mesh size", do you mean the cell number?

If you mean cell number, 5G is huge for your 16GB RAM. I think the problem might be due to lack of RAM

[klausb]

Check the mesh quality with checkMesh to see whether there's a problem.

I think the limitation is your hardware considering the problem size: no. of cores, RAM, no. of memory channels (for performance)

Maybe you have some friends with similar hardware so you could build a small cluster over a weekend to compute the problem. There are also online resources but usually not for free.

[phyczz]

Yes, I mean the cell size.
And also the size of the "polyMesh" directory.

[phyczz]

Originally Posted by hokhay

When you said "mesh size", do you mean the cell number?

If you mean cell number, 5G is huge for your 16GB RAM. I think the problem might be due to lack of RAM

Yes, the mesh size means the cell number. And also related to the size of "polyMesh"directory.

[phyczz]

Thank you for your suggestion.
I found that there was no "polyMesh"diretory if the blockMesh crashed.
As you said a cluster will be the right choice.

Thank you so much!
I will let you know if I have solve this problem.

[choist31]

Dear Physzz

I have the same error message when I try to run the parallel calculation; tutorial/incompressible/icoFoam/cavityMappingTest; I have following error message. :

Finalising parallel run
[DESKTOP-CHOIST:01362] 3 more processes have sent help message help-btl-vader.txt / cma-permission-denied
[DESKTOP-CHOIST:01362] Set MCA parameter "orte_base_help_aggregate" to 0 to see all help / error messages

Above message is generated when I use Allrun-parallel.

However, when I type decomposePar; I have a following error message:

" cannot find file "points" in directory "polyMesh" in times "0" down to constant.

I wonder if you can give me any suggestion on this error ?

Thank you in advance.

[wanghongjie]

Originally Posted by choist31

Dear Physzz

I have the same error message when I try to run the parallel calculation; tutorial/incompressible/icoFoam/cavityMappingTest; I have following error message. :

Finalising parallel run
[DESKTOP-CHOIST:01362] 3 more processes have sent help message help-btl-vader.txt / cma-permission-denied
[DESKTOP-CHOIST:01362] Set MCA parameter "orte_base_help_aggregate" to 0 to see all help / error messages

Above message is generated when I use Allrun-parallel.

However, when I type decomposePar; I have a following error message:

" cannot find file "points" in directory "polyMesh" in times "0" down to constant.

I wonder if you can give me any suggestion on this error ?

Thank you in advance.

Did you have a solution? I have the same question. Maybe you can help me.