Thank you for the reply Dr. Arndt. Let me give some more detail about the 
output with and without compress.

The code concerned snippet is the following. Just for ease in 
understanding, 'g' stands for global, 'crk' for current RK (for RK-type 
time stepping), 'avar' for auxiliary variables (three shear stresses and 
two heat fluxes) 'mu' and 'k' are viscosity and conductivity. 'gh_<vector>' 
is the ghosted version of <vector>. 'avar_vec' is an array auxiliary 
variable enums.

                // dof-wise multiplication with mu and k
        pcout << "Scaling stresses and heat fluxes\n";
        gcrk_avars[avar::txx].scale(gcrk_mu);
        pcout << "Scaled txx\n";
        gcrk_avars[avar::txy].scale(gcrk_mu);
        pcout << "Scaled txy\n";
        gcrk_avars[avar::tyy].scale(gcrk_mu);
        pcout << "Scaled tyy\n";
        gcrk_avars[avar::qx].scale(gcrk_k);
        pcout << "Scaled qx\n";
        gcrk_avars[avar::qy].scale(gcrk_k);
        pcout << "Scaled qy\n";

        // communicate
        for(avar var: avar_vec){
                pcout << "Compressing " << avar_names[var] << "\n";
                gcrk_avars[var].compress(VectorOperation::insert);
                pcout << "Communicating " << avar_names[var] << "\n";
                gh_gcrk_avars[var] = gcrk_avars[var];
        }
        pcout << "Communicated stresses and heat fluxes\n";

The output without compress (with first two lines in the last for loop 
commented) was as follows:

Scaling stresses and heat fluxes
Scaled txx
Scaled txy
Scaled tyy
Scaled qx
Scaled qy
Communicating tau_xx
Communicating tau_xy

And then it doesn't proceed. The output with compress is as follows:

Scaling stresses and heat fluxes
Scaled txx
Scaled txy
Scaled tyy
Scaled qx
Scaled qy
Compressing tau_xx
Communicating tau_xx
Compressing tau_xy

In both cases, I have verified that execution finishes (doesn't halt) when 
launched with 2 or 4 processes. I will post a minimal example if I can 
reproduce this behaviour.

Thanks

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/2a8b7a49-b976-41b5-b707-593b73350580%40googlegroups.com.

Reply via email to