Ansys: Difference between revisions

To use this approach the UDF must be compiled on an alliance cluster at least once to create the libudf subdirectory.  If you compiled your UDF on a remote system such as your laptop or a lab machine then the corresponding libudf subdirectory cannot simply be copied onto the Alliance.  That said once you have compiled your UDF on one Alliance cluster it can be transferred to another Alliance cluster without recompiling providing you load the same StdEnv environment. Then it can be simply loaded at runtime when submitting jobs by including only the second line below in your journal file, especially important when submitting parallel jobs.  Similarly if you leave both of the following lines uncommented in your journal file, the UDF will be (re)compiled automatically and then loaded each and every time you run a job on the cluster.  This would not only require additional runtime to compile the UDF unnecessarily but also impose additional complexity on every cluster since all the files required to compile your UDF would need to be present and properly setup on both/all systems.  Another way to build your UDF libudf subdirectory structure containing the <code>libudf.so</code> shared library on the Alliance would be to open your simulation in the gui on a cluster compute node (or gra-vdi) then navigate to the Compiled UDFs Dialog Box, add your UDF source file and click Build.  Before saving your cas file however verify its not defined in the Interpreted UDFs Dialog Box or the UDF Library Manager Dialog Box to ensure only the following journal file commands will control it.  To use a compiled UDF with parallel jobs it will need to be parallelized as described in the section below otherwise it will likely only work when submitted as a serial (1 core) job.