So, a solution would be calling MPI_Bcast() after every call in the if()-loop in the run()-function? Thanks!
Am Dienstag, 23. Januar 2018 14:31:58 UTC+1 schrieb Bruno Turcksin: > > Hi, > > On Tuesday, January 23, 2018 at 7:53:16 AM UTC-5, Maxi Miller wrote: > >> But now it looks like as if only the first node gets the result of the >> calculations, but the others do not, instead defaulting to the default >> values of the calculation function when not initialized. Is there a way I >> can still run the functions module.propagate_pulse() and >> module.prepare_interpolation() still only from the first node, but getting >> the result to all other nodes? Or do I have to run the function on every >> node (i.e. remove the if()-clauses)? How should I then handle the fact that >> my custom function can benefit from being run on multiple nodes at once? >> > Since you only do the calculations on one process, there is no way for the > other processes to know the results. So you have to choose, run the code > for every rank (remove the if) or use MPI_Bcast to broadcast the result of > the calculation to all the other processes. If you want to run the function > on multiple nodes, you will need to deal with raw MPI yourself. > > Best, > > Bruno > -- 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 dealii+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
