When loading a large model (which may contain many parts), prepomax can take a long time, and it also takes a long time to partition the mesh,I found that it is a serial process one by one. Can we change it to parallel processing?
Yeah, loading and meshing are still mostly serial in PrePoMax. Parallelizing could help, but it’s not trivial — might break meshing for some parts if not handled carefully.
Good news: Gmsh integration is in progress, which should speed things up once it’s fully in. Until then, large models will still take time.
Suggestion in the meantime: Try to simplify models or split assemblies into smaller, more manageable sections when possible.
I try to make features as efficient as possible. But some things will always take time. As stated, parallelizing things brings a lot of additional code that is usually hard to maintain and, in the end, does the same thing as serial code. And some things, as meshing in Netgen, will remain serial since PrePoMax uses a serial Netgen code.
What I have in mind for the future is that when meshing multiple parts at the same time, they could be meshed in parallel, each on a single thread. But then, file management and message management will give me a headache.