CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   Execution Time after few iterations very high (http://www.cfd-online.com/Forums/openfoam/106864-execution-time-after-few-iterations-very-high.html)

Tobi September 11, 2012 04:47

Execution Time after few iterations very high
 
1 Attachment(s)
Hi all,

I have a question to you. After starting my simulation (~700.000 cells on 10 cores : ~ 70.000 cells per core) the execution time of the first 6 iterations is round about 4 seconds. After that the execution time goes to very high values.

Is that normal? Depends on the cores (too many?)?
Does anyone has some advices?

Thanks tobi

wyldckat September 11, 2012 17:27

Hi Tobi,

Are you certain this is a parallel related problem? Doesn't this behavior occur when running in serial?

From your description, it's very hard to assess the reason why this might be happening. It could be due to:
  • NFS acting up, in case you're using more than one machine.
  • The number of iterations for pressure increases considerably. This could be due to various other reasons, such as mesh, fvSchemes, fvSolution....
  • Can also depend on the solver or type of simulation you're trying to perform: simpleFoam, pimpleFoam, interFoam, interDyMFoam, etc etc...
Best regards,
Bruno

Tobi September 12, 2012 03:48

Quote:

Originally Posted by wyldckat (Post 381289)
Hi Tobi,

Are you certain this is a parallel related problem? Doesn't this behavior occur when running in serial?

From your description, it's very hard to assess the reason why this might be happening. It could be due to:
  • NFS acting up, in case you're using more than one machine.
  • The number of iterations for pressure increases considerably. This could be due to various other reasons, such as mesh, fvSchemes, fvSolution....
  • Can also depend on the solver or type of simulation you're trying to perform: simpleFoam, pimpleFoam, interFoam, interDyMFoam, etc etc...
Best regards,
Bruno

Hi Bruno,
thanks for your replay. I have no experience in - how many cores are good for a cfd calculation - that the cores are working good and dont transfer data to each of them.

In that case I think its not a parallel problem :)

But thanks for your replay. I was just wondering why it is so.
Maybe it is the fact, that I start from a solution and mapped it to the new mesh (extended mesh).

Tobi


All times are GMT -4. The time now is 17:11.