Subchannel not running in parallel

[moosebuild]

Imported from https://github.inl.gov/ncrc/subchannel/issues/62 : @moose-ncrc created issue at Jun 23, 2022 08:12AM MST:

In GitLab by @tanoret on Jun 23, 2022, 09:12

Description of the enhancement

Currently, the subchannel code cannot run in parallel. The bottleneck may be at the Petsc's SNES solver. We need to make the subchannel code run in parallel.

Rationale for the enhancement

Typically, subchannel solves runs fast when compared to other TH modules. However, it would be convenient to run the solver in parallel for very large problems that may require using a fine axial mesh.

Identified impact

Code structural changes + better usability

[moosebuild]

@kyriv1980 commented at Oct 24, 2024 08:13AM MST:

It has been reported that : " If we set each individual assembly subchannel calculation as one Multiapp, the subchannel calculations cannot be parallelized and the overall calculation takes a long computational time."

Is this the same issue? I don't see why two instances of an SCM Multiapp problem wouldn't run in parallel (since they wouldn't need to communicate with eachother). Or is it that we want one problem to run on multiple processes?

Are there any input files that demonstrate those issues?

[moosebuild]

@kyriv1980 commented at Oct 24, 2024 10:42AM MST:

There are two different issues:

1. Our code is not designed to run on multiple processors. (mpi run)
2. Run multiple SCM instances in parallel (MultiApp run)