Julius,
the current MeshWorker::loop() uses the WorkStream::run() which copies in
serial.
I tried to use the WorkStream::run() version which uses colored iterators and
therefore assembles in parallel, in the sense of a tbb::parallel_for().
My code runs without using MPI as well as using an arbitrary amount of MPI
procs with a single thread each. If I use 2 MPI procs with 2 threads each, it
crashes due to a segmentation fault.
For the construction of the colored iterators I am using the
make_graph_coloring() in such a way that two cells sharing a face have
different colors.
The segmentation fault always appears while calling assemble() of the
ResidualSimple in simple.h.
Does anybody have a clue why I get a segmentation fault after adding local
data to the global dst vector inside assemble(), although I avoid write
conflicts by colored iterators?
Nothing particularly obvious comes to mind, without having tried to debug this
in more detail.
Have you tried running the program under a debugger? Segmentation faults are
often relatively easy to debug because something is accessing an invalid
address, and the debugger can tell you what object this address corresponds
to, and how this address got into the current function. Try the approach
outlined here, for example:
https://github.com/dealii/dealii/wiki/Frequently-Asked-Questions#how-do-i-debug-mpi-programs
using the `mpirun -np 4 xterm -e gdb ./my_executable` approach.
Best
W.
--
------------------------------------------------------------------------
Wolfgang Bangerth email: bange...@math.tamu.edu
www: http://www.math.tamu.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.
For more options, visit https://groups.google.com/d/optout.
