Learning Forum

[hermedo]

Hi,I have some problems in storing data (.cas and .dat files in Fluent) on FSI simulation using System Coupling that connects between Transient Structural and Fluent.nFirst of all, I'm doing my simulation in parallel mode, not serial mode.nI think my model is not large one. (7M cells for Fluent and 200K elements for Transient Structural.)nMeanwhile, after I set up the restart option in the System Coupling and executed the simulation, I confirmed that the restart files are generated right after the restart point step in Transient Structural (extension:.00x) and System Coupling (extension:.scr & .h) modules.nHowever, in case of Fluent (I think there are some problems), the time spent for storing process of .cas & .dat files is too much long. (Exactly, the extensions of .cas & .dat files are '.cas.gz' & '.dat.gz', respectively.)nAdditionally, their data capacities are about 230MB and 650MB, respectively.nI'm waiting more than 1 day to store those files.nn1. Is there any other data-storage options for parallel simulation mode compare to serial mode?n2. Is there any methods to speed up data storing process in parallel simulation mode?.If anyone wants the solver output log files, then I can provide them.nPlease leave your contacts.nTruly thanks for your reply n

[Karthik Remella]

Let's try to understand why you are seeing this issue in the first place. A 7M case should not take 1 day to save files. I'm hoping that you have sufficient memory on your system. Are you seeing similar times when you attempt to save your regular Fluent standalone files?nTo answer your questions:nNo, there is no additional storage options for parallel simulations that are different from the serial mode.nAre you using the compressed file format .gz or .h5 extension?nI hope this helps.nThanks.nKarthikn

[hermedo]

Kremella,nI really appreciate for your reply.nFirst of all, my system has sufficient memory.nThere is no any memory problem on data storing process.nAnd when I did CFD simulation with only Fluent, there was no problem just like this issue.nnTo answer your reply:nI got it. I just wondered if there are some difference between serial and parallel mode on data storing process.nYes, I'm using compressed file format. I'm taking '.cas.gz' and '.dat.gz' files from Fluent and '.h5' and '.scr' files from System Coupling.nnIs there something wrong if I use compressed file format?nThanks.n

[Rob]

Nothing wrong with the compressed format, and I assume the RAM is available. Are you saving locally or onto a network drive? n

[hermedo]

Hi, Rob.nI'm saving the result files on the server where I'm simulating the case.nI can't fully understand what 'saving locally' and 'saving onto a network drive' mean.nI'm simulating my case on the cluster (Linux OS) and also saving the above files on the same system.n

[Rob]

That's fine. We sometime see models where the compute nodes are on one machine and the disc is elsewhere. Depending on network traffic this can cause issues.n

[hermedo]

Thanks, Rob.nHowever exactly speaking, my cluster system is not just same as a personal computer that is consisted of 'one machine'.nFor more precisely, my computing system which I'm using in my simulation case have totally 8 machines (1 master node, 1 graphic node, 1 file management node, 1 storage node, and 4 computation nodes).nAnd they are connected each other with some cables for data transfer.nCan it cause the issue above?n