Hi Edscott, if you built on the standard model just run your code with mpi, e.g. mpirun -np 4 ./executable. Make sure you choose Dumux::AMGBackend as linear solver, as it is the only solver working in parallel currently (it's an AMG preconditioned BiCGSTAB). Assembly and solving will run in parallel.
There also is a parallel test in test/porousmediumflow/richards as an example using the box method. Best wishes, Timo Viele Grüße, Timo Viele Grüße, Timo Viele Grüße, Timo > Am 08.12.2017 um 21:28 schrieb Ed Scott Wilson Garcia <edsc...@imp.mx>: > > Hello all, > > So far we have optimized a low salinity flooding problem ---based on the > 2pnc model--- we are working on from a wall time of 8000 s to 1500s. And we > have noticed that the bulk of the time spent (over 90%) occurs during the > assemble of the linear system to be solved. > > Has anybody done an example parallelizing the assembler, either with threads > or MPI? > > Would this even be possible? > > Any clues, pointers or examples would be greatly appreciated. > > Kind regards, > > Edscott > > > _______________________________________________ > Dumux mailing list > Dumux@listserv.uni-stuttgart.de > https://listserv.uni-stuttgart.de/mailman/listinfo/dumux _______________________________________________ Dumux mailing list Dumux@listserv.uni-stuttgart.de https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
