CFD Online Discussion Forums - decomposePar won't break case for 8 cores

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)

- OpenFOAM Running, Solving & CFD (https://www.cfd-online.com/Forums/openfoam-solving/)

- - decomposePar won't break case for 8 cores (https://www.cfd-online.com/Forums/openfoam-solving/153223-decomposepar-wont-break-case-8-cores.html)

decomposePar won't break case for 8 cores

Quote:

numberOfSubdomains 4;

method simple;

simpleCoeffs
{
n ( 2 2 1 );
delta 0.001;
}

If I try and put in 8 cores I configure it as such:

Quote:

numberOfSubdomains 8;

method simple;

simpleCoeffs
{
n ( 4 4 1 );
delta 0.001;
}

but when decomposing the case to run it I get the following error message:

Quote:

--> FOAM FATAL ERROR:
Wrong number of processor divisions in geomDecomp:
Number of domains : 8
Wanted decomposition : (4 4 1)

From function geomDecomp::geomDecomp(const dictionary& decompositionDict)
in file geomDecomp/geomDecomp.C at line 50.

FOAM exiting

I've also tried other distribution combinations such as 4 4 2 or 4 3 1 but to no avail.

Does anyone have any idea of what I might be doing wrong?

Thanks.

Quote:

Originally Posted by Outbound (Post 547085)

Hi everyone, I have an i7 processor with 8 cores (I can view them in Ubuntu System Manager) but when I configure the decomposeParDict file to distribute it over 8 cores I get an error message that says it' an uneven distribution of cores. Right now I have it running as follows:

If I try and put in 8 cores I configure it as such:

but when decomposing the case to run it I get the following error message:

I've also tried other distribution combinations such as 4 4 2 or 4 3 1 but to no avail.

Does anyone have any idea of what I might be doing wrong?

Thanks.

hey,

n (4 4 1) gives your 16 cores, not 8...(you basically tell OpenFOAM to decompose into 16 regions when you only have 8 cores in your disposal.)

you need to divide in as (2 2 2), or (2 4 1), etc...

Thanks for the reply.

I guess I didn't understand the decomposition process.

I thought each digit meant the number of cores assigned to each direction (x y z) independently, I didn't really understand how the decomposition was accomplished by creating the mesh regions.

I'll try this again once the current simulation stops.

Once again, my thanks.