How to run parallel simulations in OpenFOAM® - Part 2
Вставка
- Опубліковано 10 лют 2025
- "How to run parallel simulations in OpenFOAM®" - Part 2
This material is published under the creative commons license CC BY-NC-SA (Attribution-NonCommercial-ShareAlike). If you plan to use it, please acknowledge it.
This video contains auxiliary material for students at the Johannes Kepler University in Linz, Austria. Students from other universities are welcome to use it for their learning purposes.
This series is based upon CFD tutorials created at the Vienna University of Technology in a cooparation with Bahram Haddadi, Christian Jordan and Michael Harasek and further improved at the Johannes Kepler University in Linz, Austria.
The used OpenFOAM version was precompiled by Andras Horvath from Rheologic GmbH, you can download it here:
rheologic.at/?q...
Very informative tutorials, thank you so much... I especially like the light humor at 7:34
complicated things are explained in very simple ways, thank you for making such knowledgeable videos
I'm still missing the difference between the "simple" and "hierarchical" methids. For cartesian decomposition, isn't the result independent of order?
Is it for distributed (multi processor) purposes where you want to keep large interface communication within the same CPU?
thank you for your work! This tutorial really help me a lot!
I am glad.
J. Nagy: "I am showing you the mesh.... you can just count them if you are bored"
me: "whaaaa? i mean hooooow? i mean why?"
honestly i love your tutorials sir but i like your comments more :D
I am glad :D
i saw your video but i have not find my solution for this problem......i have been face problem in running the "decomposePar" with "of41" while my whole work has go on "of220". bcoz here processors have not read the different zones i.e. heated,cooling zones. show saturation temperature same as maximum temperature......processors do not read the heater temperature after "mpirun -np.....-parallel"
plz...give me valuable suggestions how to resolve this ...
Seems, they have improved the scotch algorithm, now, when I ran decomposePar - the upper half was divided by vertical line on two equal parts, the lower part into two equal along the horizontal line
Thank you for your input!
That was a very good and useful tutorial
But how can I run the case in parallel mode using the openfoam ming version in Windows?
I tried this command (mpirun -np 6 pisofoam -parallel) but it didn't worked
mpiexec instead of mpirun?
@@OpenFOAMJozsefNagy thanks it worked
good and detailed tutorial series :)
Thanks Sir :)
i have 6 real core but 12 virtual threading , can i decompose in 12 parts ?
You can try, but I do not think, that it will improve the performance.
do you think 6 parallel decompositions works great ?
@@ing.nicola1706 I did not have the time to extensively test it. I would love to hear your opinion on it.
@@OpenFOAMJozsefNagy i will try ...
no, you cant divide them into 12 virtual cores since OpenFOAM does not operate like that(it will still run tough). The maximum processor core you can use is the available maximum physical core. The documentation also tells that if you are using "1 processor" then calling MPI will cause overhead and actually slows down the process. I would like to believe what he means is "1 core" instead of "1 processor" but why would anyone run mpiruns when they have only one core? I have seen parallel run makes faster in my laptop which has 1 processor(4 physical core) so i am not sure if it is a comment made for older versions. You can find nice highlight in here wiki.openfoam.com/Parallelization_by_Joel_Guerrero
, thank you so much for very informative tutorials. I have an error message while I type decomposePar. I wounder if I can have other contact e-mail regarding my error message to you. Thank you in advance.