Scotch decomposition error
Hello everyone,
I've been trying to decompose my 2-D Mesh using scotch. Unfortunately every cell gets distributed to the same processor 0. The remaining 3 processors wont get any cells. Did this happen to anyone else? I'd appreciate any kind of help. :) - Mirko Code:
Calculating distribution of cells Code:
numberOfSubdomains 4; |
Hi,
I assume you didn't get a reply because there is not enough information to see an "error" in your input files... Did you solve it? Otherwise you may want to share your full decomposeParDict... |
Hi,
I found the same error in OpenFOAM-7. This is my decomposeParDict: Code:
/*--------------------------------*- C++ -*----------------------------------*\ Code:
Decomposing mesh region0 Have you found any solution? |
Here's what my decomposeParDict looks like for 16 cores and it runs fine:
Code:
numberOfSubdomains 16; |
Hi guys, did you solve that issue? I'm getting the same errors in my decomposition. Never had that before while I used the ESI-Versions (v1906, v1912, v2006) of OF. Now that I changed to the foundation version (v7) I get this error:
"Processor 47: field transfer --> FOAM Warning : From function Foam::polyMesh::polyMesh(const Foam::IOobject&) in file meshes/polyMesh/polyMesh.C at line 332 no points in mesh" ...for multiple processors in my decomposePar-log. I checked my whole geometry many times, but I couldn't see anything getting wrong. Does anyone have any suggestions or already solved that issue? I just used method simple for decomposition of a 2D-mesh on 48 cores with simpleCoeffs: n (4 12 1) and delta 0.001. |
I've seen errors happen for simple/hierarchical if you have much more cells in the center of your domain as on the edges, but I haven't identified the source of the more general 0 cells problem. You can try the tips of Mark Olesen: https://www.cfd-online.com/Forums/op...tml#post794214
|
Hello Louis,
thanks for your fast reply! I think i just managed this issue. Somehow the cht-Solver of the foundation-version has a problem by using simple method. I just needed to change to hierarchial, which is almost the same. I couldn't imagine before, that simple method was the problem, because it is the most common used and easiest method. Well.. changing versions is not that trivial anymore it seems :D. But thank you again for your advice :). |
All times are GMT -4. The time now is 17:03. |