I am trying to run the step-32 example program on a cluster using deal.ii 7.1.0 and Trilinos 10.4.2. It runs fine (if quite slowly) on a single processor, but if I try to give it more processes, it hangs for all intents and purposes (the vast majority of the time being spent in the BoussinesqFlowProblem::solve() step). I believe that the problem comes down to something spawning way too many threads on the compute nodes, which are then interfering with each other. I am not sure why this is the case, as I have deal.ii compiled with --disable-threads. Is it possible that these are being spawned by the Trilinos libs?
Not really -- Trilinos doesn't create threads. What indications do you have that threads are the problem? And can you reproduce this problem with the current deal.II svn version?
I've seen such problems before if the number of cells is small, but I've never really been able to figure out what is happening and it was never really reproducible. The deal.II FAQ has an entry on how to find out what's going on in case an MPI process appears to hang. Can you try to follow these suggestions?
Best W. ------------------------------------------------------------------------ Wolfgang Bangerth email: [email protected] www: http://www.math.tamu.edu/~bangerth/ _______________________________________________ dealii mailing list http://poisson.dealii.org/mailman/listinfo/dealii
