Matthew Knepley

Can the load balance problem be phrased as just graph partitioning with a balance constraint (as PArMetis supports)? (edited by @drwells to remove extra email text)

Also, by checking the refinement criteria you input, you can force p4est to deliver blocks if you want, and the 2:1 balance is also optional.

I have been using it for a while now. All the plasma stuff I am doing with Mark Adams uses p4est. You can convert seamlessly to a Plex (and pretty...

Are you using GPU-aware MPI? We have seen several unexplained segfaults with this that vanish using the normal build of MPI. So far, the implementors have not been motivated to...

Yes

@baagaard-usgs I have pushed to `knepley-pylith`. Now there is no failure, but I am not sure we are getting the right data out. How do I check?

@baagaard-usgs There were many things wrong. I have pushed to `knepely/pylith`. Let's try it again at the meeting.

By default, ASM has an algebraic overlap of 1.

@boyceg The main difference I see is that PETSc does a 2-stage communication. First we communicate the global RHS to a local RHS, with the whole halo. Next we restrict...

@baagaard-usgs I think this is fixed. I have force pushed the PETSc branch. The problem was that the residual is cloned from the solution, and clone did not copy the...