Hello, Sir,

For the reason the simulation will take a long time in a common computer, I have to run the simulation on a super computer.

ssh2 super_computer_name ; Fluent

My question is whether I can run the program on the background of the super computer. Then I logouted my machine. After several hours, I login my machine to take a look at the program. How can I get the results from background of the supercomputers to foreground of my computer?

Thanks very much.

Jie

You can do this by running a batch file on the mainframe computer. If you look in the user manual, there are instructions for writing and running batch files, and using text interface, depending on the type of computer system fluent is on. For me, ours is on a unix machine, so that is what I know. Would you like to see a sample?

Typically, I would have FLUENT write out a results file every 50 to 100 iteration (under write>auto-save ??). Also dump everything that would go to the screen to a file using something like run.exe > out.log. While your job is running (on the supercomputer) copy the latest result and log files from the computer's hard drive (scratch space) to your local or working directory. This methods will allow your job to keep running. It has been a while since I have worked with on an SGI Origin array, so I cannot remember the exact commands. As Cindy mentioned put everything in a text script file.

Yes, Can you give me an example? My machine is also a unix machine. Thanks.

Hello,

I set auto-save to 10 interations. But the problem said that save errors existed.

Jie

Here is and example of an input text file for a case name of 3Dtank.cas, which I have called 3Dtankin.txt:

-

file/read-case-data 3Dtank.cas

solve/monitors/residual/plot no

solve/monitors/residual/print yes

it 50

wd 3Dtank.dat

yes

it 50

wd 3Dtank.dat

yes

exit

-

I run it in the background from the command prompt on the unix machine using the following:

fluent 3d -gu <3Dtankin.txt>& 3Dtankou.txt &

I chose the output file name to be 3Dtankout.txt. The -gu in the command line will make sure it doesn't try to turn on the graphics. It doesn't like to do that in batch mode, so you would get error messages. I make sure the residual plotting is off, since that needs graphics, and turn on printing so the output file has something I can see. You can put other commands in your text file depending on what you want it to do. If you are familiar enough with the menus, the text commands pretty much follow the same structure, with a "/" between subdirectories. It also recognizes shortcuts. wd stands for write-data, it stands for iterate. The "yes' is to answer the question fluent would have at that point, which is "3Dtank.dat already exists. OK to overwrite?" Let me know how it goes.

Thank you very much. Let me try.

it 50

wd 3Dtank.dat

yes

it 50

wd 3Dtank.dat

yes

Is '50' the times of interations for saving files? Thanks very much.

------Jie

hello,

If I want to run for 5000 interations and save data every 100 interations. How can I write the code.

Thanks very much.

---Jie

You need to set it in File/Autosave, put 100 for the case or data file or both, then save the case file and the rest is the same: it 5000

In UNIX, you need to make sure the file name you put in the Autosave filename field has the correct path. If you only put a filename there it will use the default/current path to save the file.

You could say

it 100

wd outputfilename.txt

yes

and do that 50 times. If I have that many iterations to do, I usually iterate longer between saves, like maybe 200 or 300. I have not had much luck myself with the autosave feature, but this works well enough for me. Everytime you say wd it is saving the data.