Continue OpenFoam parallel run with different number of processors

orbitaljet · August 15, 2021, 07:33

Hi all. I am running a large case on a cluster. I have a job completed on 16 cores up to a timestep of 10000s. I have reconstructed the data using reconstructPar and deleted all the processor directories (processor0, processor1...).

How do I continue the job to a timestep of 20000s using 32 cores instead?

This is my script:

Code:

decomposePar -force
mpirun -np 16 renumberMesh -overwrite -parallel
mpirun -np 16 overPimpleDyMFoam -parallel | tee log.solver

I have edited the controlDict to start from the latestTime and the decomposeParDict has been changed for 32 processors now. After the decomposePar command, do I need to renumber the mesh again with 32 cores before running the solver itself?

Tobermory · August 16, 2021, 04:53

From a quick look at the source code (https://github.com/OpenFOAM/OpenFOAM...renumberMesh.C), I don't see any reference to the number of processors in the run in the coding, so I don't think that the way you have decomposed the problem matters. Instead, the utility is used to make the LDU matrix addressing more efficient - see eg this post.

orbitaljet · August 17, 2021, 00:18

Great, just needed some confirmation before submitting the job. Thank you!

August 15, 2021, 07:33	Continue OpenFoam parallel run with different number of processors	#1
orbitaljet New Member Join Date: Aug 2021 Posts: 5 Rep Power: 6	Hi all. I am running a large case on a cluster. I have a job completed on 16 cores up to a timestep of 10000s. I have reconstructed the data using reconstructPar and deleted all the processor directories (processor0, processor1...). How do I continue the job to a timestep of 20000s using 32 cores instead? This is my script: Code: decomposePar -force mpirun -np 16 renumberMesh -overwrite -parallel mpirun -np 16 overPimpleDyMFoam -parallel \| tee log.solver I have edited the controlDict to start from the latestTime and the decomposeParDict has been changed for 32 processors now. After the decomposePar command, do I need to renumber the mesh again with 32 cores before running the solver itself?

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Decomposing meshes	Tobi	OpenFOAM Pre-Processing	22	February 24, 2023 10:23
GenerateVolumeMesh Error - Surface Wrapper Self Interacting (?)	AndreP	STAR-CCM+	10	August 2, 2018 08:48
[OpenFOAM.org] OF2.3.1 + OS13.2 - Trying to use the dummy Pstream library	aylalisa	OpenFOAM Installation	23	June 15, 2015 15:49
Can not run OpenFOAM in parallel in clusters, help!	ripperjack	OpenFOAM Running, Solving & CFD	5	May 6, 2014 16:25
DecomposePar unequal number of shared faces	maka	OpenFOAM Pre-Processing	6	August 12, 2010 10:01

August 16, 2021, 04:53		#2
Tobermory Senior Member Join Date: Apr 2020 Location: UK Posts: 859 Rep Power: 19	From a quick look at the source code (https://github.com/OpenFOAM/OpenFOAM...renumberMesh.C), I don't see any reference to the number of processors in the run in the coding, so I don't think that the way you have decomposed the problem matters. Instead, the utility is used to make the LDU matrix addressing more efficient - see eg this post.

August 17, 2021, 00:18		#3
orbitaljet New Member Join Date: Aug 2021 Posts: 5 Rep Power: 6	Great, just needed some confirmation before submitting the job. Thank you!