Re: [deal.II] Re: Error with Petsc and and multithreading?
On 9/9/19 3:10 PM, Konrad wrote: > > Thanks a lot, Wolfgang! The error message is actually quite clear. What > I found out in addition is that it is advantageous and helpful to read > the documentation ... ;-) Yes, no doubt :-) We're spending a lot of time writing it ;-) Cheers W. -- Wolfgang Bangerth email: bange...@colostate.edu www: http://www.math.colostate.edu/~bangerth/ -- 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. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/35630390-2478-266e-516f-6cbacabaaa16%40colostate.edu.
Re: [deal.II] Re: Error with Petsc and and multithreading?
> > Yes, this is indeed the case, and is what the error message was trying to > suggest. If you think that the error message could have been clearer, feel > free to propose a better wording and we'll include it for the next > release! > Thanks a lot, Wolfgang! The error message is actually quite clear. What I found out in addition is that it is advantageous and helpful to read the documentation ... ;-) Best , Konrad -- 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. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/6aa4d4eb-f91c-4519-a707-4833225fe4d8%40googlegroups.com.
Re: [deal.II] Re: Error with Petsc and and multithreading?
On 9/9/19 2:47 AM, Konrad wrote: > OK, just to answer the question: If you are running with petsc you should not > call > > dealii::Utilities::MPI::MPI_InitFinalize > mpi_initialization(argc, argv, dealii::numbers::invalid_unsigned_int); > > since this will invoke threading if you start a number of mpi processes that > is not equal the number of cores you use. Instead one should pass (when using > generic linear algebra) > > #ifdef USE_PETSC_LA > dealii::Utilities::MPI::MPI_InitFinalize > mpi_initialization(argc, argv, /* disable threading for petsc */ 1); > #else > dealii::Utilities::MPI::MPI_InitFinalize > mpi_initialization(argc, argv, dealii::numbers::invalid_unsigned_int); > #endif > > I guess that was the bug. Yes, this is indeed the case, and is what the error message was trying to suggest. If you think that the error message could have been clearer, feel free to propose a better wording and we'll include it for the next release! Best W. -- Wolfgang Bangerth email: bange...@colostate.edu www: http://www.math.colostate.edu/~bangerth/ -- 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. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/919b1577-e0eb-5e43-664d-82de1c8d3ab4%40colostate.edu.
[deal.II] Re: Error with Petsc and and multithreading?
OK, just to answer the question: If you are running with petsc you should not call dealii::Utilities::MPI::MPI_InitFinalize mpi_initialization(argc, argv, dealii::numbers::invalid_unsigned_int); since this will invoke threading if you start a number of mpi processes that is not equal the number of cores you use. Instead one should pass (when using generic linear algebra) #ifdef USE_PETSC_LA dealii::Utilities::MPI::MPI_InitFinalize mpi_initialization(argc, argv, /* disable threading for petsc */ 1); #else dealii::Utilities::MPI::MPI_InitFinalize mpi_initialization(argc, argv, dealii::numbers::invalid_unsigned_int); #endif I guess that was the bug. Best, Konrad -- 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. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/628d94ee-9e38-4dfd-b0b7-0a84d7b832ac%40googlegroups.com.