Compile and load a UDF automatically when running Fluent on an HPC cluster

[a.y.al-wardi]

I have a transient Fluent simulation with a compiled UDF (`decomp_fff1.c`) that has been successfully compiled and loaded on my Windows workstation. The UDF function is `ozone_decomposition_source` and is assigned to the appropriate cell zone.

 

I now want to run the same case on an HPC cluster (Eddie), where Fluent runs on Linux.

 

I understand that the Windows `libudf` should not be transferred to the Linux cluster and that the `.c` source file should instead be compiled on Eddie to generate a Linux-compatible `libudf`.

 

What is the recommended way to include the UDF compilation/loading in a Fluent journal (`.jou`) so that the UDF is automatically compiled and loaded when the case is submitted as a non-interactive HPC job?**

 

I have recorded the compilation process on Windows, but Fluent generates `cx-gui-do` commands such as: `(cx-gui-do cx-activate-item "Compiled UDFs*Frame8*PushButton2(Build)")` rather than a simple TUI command.

 

Is there a recommended TUI/journal command for compiling a `.c` UDF on Linux?

 

Thank you!

 

[Rob]

Use a TUI (console) journal. Click into the console and press to open a menu type it in and hit    To move up a menu level use q (Q).  The Search tool in the top right is also helpful for finding things. 

The command is likely under /define/user-defined   and always use the full command preceded by the /  

 

[abhiroy]

Hello, 

You can use the following command to compile the UDF 

/define/user-defined/compiled-functions compile libudf Yes sample_udf.c ""

 

Then use following command to load the library 

/define/user-defined/compiled-functions load libudf