Alan Williams

@sayerhs that's interesting. In general, are we instantiating more than we need to? For debugging a particular case, can you still link if you only instantiate for the element type...

This doesn't look like a nalu-wind issue. Is there a reason it's filed as a nalu-wind issue rather than a kokkos and/or tpetra issue?

I'm currently working on kernel/ScalarMassElemKernel. (It's done and passing regression tests, but I don't have it unit-tested yet.)

A note for when we return to element-based GPU conversion: Regarding AssembleFaceElemSolverAlgorithm. The run_algorithm method is converted and unit-tested, but the execute method needs to be converted next. Need to...

Why put in such a large test case? Test cases should contain relevant features and complexities, but not be huge in size.

This seems like something we can debug locally which should be a lot easier than debugging on crusher. It seems possible that it is the same bug.

I think overall it's a good idea, and it's going in the right direction. I think a big benefit of reducing dependencies on Realm, is that it would be easier...

@psakievich Yes that's an interesting idea. I might need to think through it a little more, but I like it.

ok that seems reasonable, looks good to me.

ok we'll get a story on the backlog to take care of this.