Performance 1.5-dev with ggi

bastil · June 30, 2009, 11:51

Dear all,

I run a large case with about 30 million cells on 32 CPUs using openmpi1.6 from OpenFOAM build with mx-support. One iteration for this run takes about 80 seconds. Fluent uses about 60 seconds with the coupled solver, the seggregated solver is much quicker. My FLUENT setup is optimised, my FOAM setup is not (I do not really know how to do). This difference looks extremely large to me. What can this be caused by?

Thanks BastiL

bastil · June 30, 2009, 15:46

Some more infos to add:
FLUENT runs with HP-MPI and 2nd order upwind. OF with the central differencing schemes. I have read that these are quickest in OF.
I am not sure if the poor performance is caused by the ggis - I could try run a comparison case without ggi.

Regards BastiL

June 30, 2009, 11:51	Performance 1.5-dev with ggi	#1
bastil Senior Member BastiL Join Date: Mar 2009 Posts: 530 Rep Power: 20	Dear all, I run a large case with about 30 million cells on 32 CPUs using openmpi1.6 from OpenFOAM build with mx-support. One iteration for this run takes about 80 seconds. Fluent uses about 60 seconds with the coupled solver, the seggregated solver is much quicker. My FLUENT setup is optimised, my FOAM setup is not (I do not really know how to do). This difference looks extremely large to me. What can this be caused by? Thanks BastiL

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Performance of GGI case in parallel	hannes	OpenFOAM Running, Solving & CFD	26	August 3, 2011 03:07
GGI ERCOFTAC and general questions	david	OpenFOAM Running, Solving & CFD	8	August 25, 2008 09:22
OpenFOAM Version 1.5 Released	OpenFOAM discussion board administrator	OpenFOAM Announcements from ESI-OpenCFD	0	July 14, 2008 18:22
P4 1.5 or Dual P3 800EB on Gibabyte board	Danial	FLUENT	4	September 12, 2001 11:44
CFX4.3 -build analysis form	Chie Min	CFX	5	July 12, 2001 23:19

June 30, 2009, 15:46		#2
bastil Senior Member BastiL Join Date: Mar 2009 Posts: 530 Rep Power: 20	Some more infos to add: FLUENT runs with HP-MPI and 2nd order upwind. OF with the central differencing schemes. I have read that these are quickest in OF. I am not sure if the poor performance is caused by the ggis - I could try run a comparison case without ggi. Regards BastiL