saurabh636

Hi Denis Thanks, i have seen that example, but i am looking for running shared memory parallell not mpi. Anything? Thanks Saurabh On Sun, Jun 20, 2021, 11:52 PM Denis...

My apologies, the situation is indeed a bit complicated, what I have is a domain decomposition setup for FE, but it does not directly rely on MPI, instead it is...

Unfortunately, I cannot access MPI communicator, but I will give re-implementation using native API a shot.

Hi For AMGCLs parallel solver, examples suggest using the AMGCL's distributed matrix `auto A = std::make_shared(comm, local, remote);` where local and remote matrices are "crs" types. Typically, in FE setting,...

Does a row need to be entirely on a single proc? or could it be split across procs as well. My matrix adapter is based on CRS as well which...

OK, I can ensure the assembly perhaps, however, not sure if it will be efficient to gather the entire row on a single processor, this partition naturally occurs in a...

If I am able to presplit the matrix into local and remote parts, do I still need to ensure a contiguous chunk of rows belong entirely to a single processor?...