Execution time problem after first write interval

sean_mcintyre · July 25, 2008, 15:46

Hi,
I am currently running OpenFOAM 1.4.1, and have been experiencing a problem with the solvers (currently rasInterFoam, though it has shown up with other solvers as well). When running parallel jobs, the execution time increases from about 15 seconds per time step to about 10000 seconds after the first write interval, and stays that way. Upon upgrading to 1.5, we still experienced the problem, which doesn't happen in single processor jobs. In 1.4.1-dev, it was less notable, only about doubling the execution time. Has anyone else been experiencing the same problems? If so, do you have any idea what's the cause? I have used both LAM and OpenMPI, and had the same problem.

Thanks,
Sean McIntyre

sylvester · July 28, 2008, 05:36

Hi Sean,

We've had the same problem. It was related to an NFS bug in the RedHat kernel. I don't know which version, but updating the kernel solved our problem.

Sylvester

eugene · July 28, 2008, 06:12

I can confirm this. I think it was RHEL 4 update 3. We never found out why it happened only with OpenFOAM though.

sean_mcintyre · July 28, 2008, 09:46

Thank you so much! I'll have to see if our HPC person can update the kernel for us. We wanted to just blame LAM, but it happened with the MPI too. Thanks again,

Sean McIntyre

July 25, 2008, 15:46	Hi, I am currently running Op	#1
sean_mcintyre New Member Sean McIntyre Join Date: Mar 2009 Location: State College, PA Posts: 11 Rep Power: 17	Hi, I am currently running OpenFOAM 1.4.1, and have been experiencing a problem with the solvers (currently rasInterFoam, though it has shown up with other solvers as well). When running parallel jobs, the execution time increases from about 15 seconds per time step to about 10000 seconds after the first write interval, and stays that way. Upon upgrading to 1.5, we still experienced the problem, which doesn't happen in single processor jobs. In 1.4.1-dev, it was less notable, only about doubling the execution time. Has anyone else been experiencing the same problems? If so, do you have any idea what's the cause? I have used both LAM and OpenMPI, and had the same problem. Thanks, Sean McIntyre

July 28, 2008, 05:36	Hi Sean, We've had the same	#2
sylvester Member Roland Join Date: Mar 2009 Location: Netherlands Posts: 92 Rep Power: 17	Hi Sean, We've had the same problem. It was related to an NFS bug in the RedHat kernel. I don't know which version, but updating the kernel solved our problem. Sylvester

July 28, 2008, 06:12	I can confirm this. I think it	#3
eugene Senior Member Eugene de Villiers Join Date: Mar 2009 Posts: 725 Rep Power: 21	I can confirm this. I think it was RHEL 4 update 3. We never found out why it happened only with OpenFOAM though.

July 28, 2008, 09:46	Thank you so much! I'll have	#4
sean_mcintyre New Member Sean McIntyre Join Date: Mar 2009 Location: State College, PA Posts: 11 Rep Power: 17	Thank you so much! I'll have to see if our HPC person can update the kernel for us. We wanted to just blame LAM, but it happened with the MPI too. Thanks again, Sean McIntyre

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Is there a way to write the time step size, time a	may	FLUENT	6	November 22, 2009 12:52
why did my time interval set for output nouse?	ShFlow	FLOW-3D	2	January 11, 2009 21:25
Reading case file at certain time interval	Chin Fook	FLUENT	1	September 11, 2008 07:54
Measuring an Execution Time - f90	su	Main CFD Forum	11	July 4, 2008 08:21
Execution problem	Jiri	Siemens	3	October 16, 2003 10:46