Issues with Parallelizing DEFINE_GRID_MOTION UDF
Hello everyone,
I am working on parallelizing my DEFINE_GRID_MOTION UDF and would appreciate any guidance. My simulation runs correctly in serial mode, but I am encountering issues when executing the UDF in parallel.
To debug, I used the Display Zone Motion feature to identify where the code encounters fatal errors. After some trys, the UDF runs perfectly in Display Zone Motion, but it fails when using Preview Mesh Motion.
I am trying to understand why the code behaves differently in these two cases. I suspect this may be related to the synchronization of compute nodes (I used PRF_GSYNC()). Specifically, I would like to clarify whether DEFINE_GRID_MOTION updates the mesh nodes for all compute nodes before applying remeshing and dynamic mesh algorithms, or if it updates each compute node individually before merging the results.
Additionally, I would like to know whether n_udmi can be accessed through a dynamic pointer within DEFINE_GRID_MOTION, or if it is only feasible through the domain pointer.
Any insights or suggestions regarding DEFINE_GRID_MOTION in parallel simulations would be greatly appreciated!
Thank you in advance for your help.
Best regards,
Dimitris