CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Community Contributions (https://www.cfd-online.com/Forums/openfoam-community-contributions/)
-   -   [ImmersedBoundary] mpirun gives error despite I can run my case serially (https://www.cfd-online.com/Forums/openfoam-community-contributions/207103-mpirun-gives-error-despite-i-can-run-my-case-serially.html)

ordinary September 24, 2018 16:13

mpirun gives error despite I can run my case serially
 
2 Attachment(s)
Hello,

I have a 2D case which consists of 11496 cells. Unfortunately I can only run it serially. Whenever I try to type
Code:

decomposePar

mpirun -np 8 icoDyMIbFoam -parallel

it gives me error: Attachment 65725

Here is my decomposeParDict file in addition: Attachment 65726

I tried to find a solution and encountered a post mentions swap configuration. https://www.cfd-online.com/Forums/op...-problems.html
But I don't think my problem is related with swap memory. The laptop I use has 16 GB RAM and 7.6 GB swap size (swapon command says so). I even tried to run the case in an HPC cluster.

To summarize, I can't run the case in parallel but can run it serially.

Regards,

Emre

wyldckat October 7, 2018 08:19

Quick answer:
  1. Dividing so very few cells by 8 cores, gives an average of 1437 cells.
  2. The error message states the following:
    Code:

    Signal: Floating point exception (8)
    libfoam.so(_ZN4Foam17DICPreconditioner15calcReciprocalDERNS_5FieldIdEERKNS_9lduMatrixE+0x107

  3. I know it's hard to spot the detail, but there was a problem with a mathematical operation when running the method "DICPreconditioner::calcReciprocal".
  4. This is very likely to be related to the very few cells being used for each core, because it's likely missing the information that is within the other processors.
  5. The solution may be to either not run in parallel or only run in parallel with 2 cores.
  6. Furthermore, given this information:
    Code:

    [3] Number of IB cells: 0
    [7] Number of IB cells: 0
    [2] Number of IB cells: 0
    [6] Number of IB cells: 0
    [5] Number of IB cells: 32
    External flow
    [4] Number of IB cells: 29
    [0] Number of IB cells: 31
    [1] Number of IB cells: 31

    I suspect that the problem is due to having the IB surface being shared between too many processors... it may be possible that it will only work if the IB surface is only present within a single processor domain.

ordinary October 7, 2018 15:39

Thank you very much Mr. Santos. I decreased simpleCoeffs( ) as you suggested. Now it solves faster.

Regards,
Emre

zjz November 6, 2018 08:19

parallel with icoDyMIbFoam
 
Quote:

Originally Posted by ordinary (Post 707300)
Hello,

I have a 2D case which consists of 11496 cells. Unfortunately I can only run it serially. Whenever I try to type
Code:

decomposePar

mpirun -np 8 icoDyMIbFoam -parallel

it gives me error: Attachment 65725

Here is my decomposeParDict file in addition: Attachment 65726

I tried to find a solution and encountered a post mentions swap configuration. https://www.cfd-online.com/Forums/op...-problems.html
But I don't think my problem is related with swap memory. The laptop I use has 16 GB RAM and 7.6 GB swap size (swapon command says so). I even tried to run the case in an HPC cluster.

To summarize, I can't run the case in parallel but can run it serially.

Regards,

Emre

Hi,
i run sample pitchingPlate in foam-extend4.0 with icoDyMIbFoam in parallel,
but it give me errorUsage: icoDyMIbFoam [-DebugSwitches key1=val1,key2=val2,...] [-DimensionedConstants key1=val1,key2=val2,...] [-InfoSwitches key1=val1,key2=val2,...] [-OptimisationSwitches key1=val1,key2=val2,...] [-Tolerances key1=val1,key2=val2,...] [-case dir] [-dumpControlSwitches] [-noFunctionObjects] [-parallel] [-help] [-doc] [-srcDoc]


Usage: icoDyMIbFoam [-DebugSwitches key1=val1,key2=val2,...] [-DimensionedConstants key1=val1,key2=val2,...] [-InfoSwitches key1=val1,key2=val2,...] [-OptimisationSwitches key1=val1,key2=val2,...] [-Tolerances key1=val1,key2=val2,...] [-case dir] [-dumpControlSwitches] [-noFunctionObjects] [-parallel] [-help] [-doc] [-srcDoc]



--> FOAM FATAL ERROR:
Wrong number of arguments, expected 0 found 1


FOAM exiting



--> FOAM FATAL ERROR:
Wrong number of arguments, expected 0 found 1


FOAM exiting

-------------------------------------------------------
Primary job terminated normally, but 1 process returned
a non-zero exit code.. Per user-direction, the job has been aborted.
-------------------------------------------------------
--------------------------------------------------------------------------
mpirun detected that one or more processes exited with non-zero status, thus causing

How do you run the sample in parallel?
thank you.

ordinary November 6, 2018 08:38

Quote:

Originally Posted by zjz (Post 714320)
How do you run the sample in parallel?
thank you.

Hello. I run it via these commands:
Code:

decomposePar
potentialIbFoam
 mpirun -np 8 icoDyMIbFoam -parallel

8 in the code above means CPU cores. It needs to be same with decomposeParDict file in constant folder.

Regards,

Emre


All times are GMT -4. The time now is 05:55.