[hsieh]

Hi,

I have a customized solver (not a standard OpenFOAM solver). It ran fine in serial. When I ran it in parallel, AMG failed. Changed to ICCG/BICCG, ran with only a warning at each time step. I have not checked whether the results are correct.

When looking at the code, is there a way to tell whether it is "parallel-ready"?

Pei

[jojo]

Hi,

I have a customized solver as well. However, I do not manage to run it in parallel while it runs fine in serial.

I get the following message:
-----------------------------------------------------------------------------
It seems that [at least] one of the processes that was started with
mpirun did not invoke MPI_INIT before quitting (it is possible that
more than one process did not invoke MPI_INIT -- mpirun was only
notified of the first one, which was on node n0).

mpirun can *only* be used with MPI programs (i.e., programs that
invoke MPI_INIT and MPI_FINALIZE). You can use the "lamexec" program
to run non-MPI programs over the lambooted nodes.
-----------------------------------------------------------------------------

It seems that wmake compilation of custom solvers has an issue with parallelization (I have seen another post with the same problem).

[hartinger]

Hey,

I think first you would need to understand the existing solvers like GaussSeidel. Set up a very simple test case like 10x1x1 cells and a very simple equation and check whether each processor receives what you would expect.

markus

[mattijs]

- make sure your solver uses argList to extract the command line options
- make sure your solver gets called with a -parallel option