CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (http://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Parallel computation with OF141 (http://www.cfd-online.com/Forums/openfoam-solving/59502-parallel-computation-of141.html)

benru August 10, 2007 10:01

Now we tested ordinary & paral
 
Now we tested ordinary & parallel calculations with bubbleFoam of OF-1.4.1. Region was decomposed on two part for two core procesor (Intel Pentium E6700), and result was very strange : solving time was about 7 minutes for two cores, but 5 minutes without decomposition ('top' showed 100%-loading of both cores for parallel case but only single core 100%-loading for ordinary case). Could anybody explain, what is the matter ?

Best regards
ruben

gschaider August 15, 2007 05:33

How many cells does your case
 
How many cells does your case have? If there are to few the processors spend more time coordinating their efforts than calculating and that is when you see that kind of behaviour.

benru August 15, 2007 08:49

It was bubbleColumn case (from
 
It was bubbleColumn case (from tutorials) for bubbleFoam solver, 25x75x1=1875 cells. I think, it would not cause this problem, doesn't it ?

ruben

fra76 August 15, 2007 09:21

Yes, that's the reason. You n
 
Yes, that's the reason.
You need much bigger cases to take advantage of parallel computation...


All times are GMT -4. The time now is 23:19.