|
[Sponsors] |
July 24, 2014, 10:07 |
Parallelization
|
#1 |
Member
Join Date: Oct 2011
Location: Thessaloniki, Greece
Posts: 75
Rep Power: 15 |
Hello,
I wish to run a computation in parallel. However, I wish to defined by myself how to partionate my geometry. As for example, region 1 should be the partition 0, solved by CPU1 region 2 should be the partition 1, solved by CPU2 region 3 should be the partition 2, solved by CPU3 Does anyone if this is possible? When I go throught the GUI to Solvers>partitioning, I can choose between partioning by regions or by continua. However, doing so does not really help. (it seems that the automatic partioning tried to keep the same number of cells by partions, while my regions have different number of cells) Thanks a lot for any help |
|
July 24, 2014, 22:48 |
|
#2 |
Senior Member
Join Date: Nov 2010
Location: USA
Posts: 1,232
Rep Power: 25 |
This is not possible. You can partition by region which will try and get what you want, but that's it. If each region is a separate continuum you can try that.
I don't really see a reason why this should be necessary though. |
|
July 25, 2014, 06:43 |
|
#3 |
Member
Join Date: Oct 2011
Location: Thessaloniki, Greece
Posts: 75
Rep Power: 15 |
Thanks a lot for your answer.
I did try the partioning by region or by continuum. However, I think StarCCM+ mostly focuses on getting the same number of cells by partition, as it anyway devide my regions to fit that rule. I am currently coupling a StarCCM+ computation to an external code. Only one region of the whole geometry is solved by the external code. If this region is divided, I am getting trouble in the communication between the 2 codes (made with field functions) and the computation crashes. I was doing the same in StarCD, where I could specify a set of cell and assigned it by CPU. I cannot yet do the same using the regions or cell set in StarCCM+. I am wondering if I can access to the partitioning strategy via java macro. I will investigate that way too. |
|
Tags |
parallel computation |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Automatic parallelization tools on SMPs | lalupp | Hardware | 0 | June 25, 2012 11:54 |
MPI Aborts when using reactingFoam with cyclic BC in parallelization | jose_rodrig | OpenFOAM | 1 | February 14, 2011 04:05 |
Surface interpolation schemes and parallelization | jutta | OpenFOAM Running, Solving & CFD | 0 | February 25, 2010 15:32 |
parallelization calculation | SARU | OpenFOAM | 0 | December 15, 2009 20:24 |
Parallelization of Riemann ??? I need Roe or Osher | Rouboa | Main CFD Forum | 0 | July 19, 1999 15:26 |