PETSc: Segmentation Violation
Hello,
thanks for the efforts on SfePy and even adding a parallel solver. Unfortunately, the parallel examples don't work for me. I've tried with petsc4py 3.6.0 and PETSc 3.6.0 as well as 3.6.1.
I appned the error message from the command "python examples/multi_physics/biot_parallel_interactive.py output-parallel". When run in parallel the message is basically the same, but multiple times.
Best regards
Hi,
On 07/28/2015 03:32 PM, derj...@web.de wrote:
Hello,
thanks for the efforts on SfePy and even adding a parallel solver.
Note that it is still in a highly experimental state.
Unfortunately, the parallel examples don't work for me. I've tried with petsc4py 3.6.0 and PETSc 3.6.0 as well as 3.6.1.
Thanks for the report!
Could you try petsc 3.4.2 (petsc4py 3.4)? Those are the versions I have tested the code with. It might be that something changed in the API between 3.4 and 3.6.
I appned the error message from the command "python examples/multi_physics/biot_parallel_interactive.py output-parallel". When run in parallel the message is basically the same, but multiple times.
I will try to debug it, but cannot say when now (other work to do...). Could you try the suggestions mentioned in the error message?
[0]PETSC ERROR: Try option -start_in_debugger or -on_error_attach_debugger [0]PETSC ERROR: or see http://www.mcs.anl.gov/petsc/documentation/faq.html#valgrind [0]PETSC ERROR: or try http://valgrind.org on GNU/linux and Apple Mac OS X to find memory
- it could help to pinpoint the line in sfepy/parallel/parallel.py, where the error occurs.
r.
Hello Robert,
Note that it is still in a highly experimental state. Yes, yet it is very appreciated.
Thank you very much for your help. It turned out, the mistake was completely on my side. mpi4py used my "native" mpicc instead of the one coming with PETSc.
Sorry for bothering you with this, instead of looking deeper into this myself. But I never used gdb before and it was quite some effort now, to make it work.
So, bottom line, if anyone with a similar error ever reads this. Make sure that mpi4py uses the correct mpicc coming with PETSc. E.g.:
env MPICC=/path/to/mpicc pip install mpi4py
best regards, JonnyB
On 07/28/2015 08:09 PM, derj...@web.de wrote:
Hello Robert,
Note that it is still in a highly experimental state. Yes, yet it is very appreciated.
Thank you very much for your help. It turned out, the mistake was completely on my side. mpi4py used my "native" mpicc instead of the one coming with PETSc.
Sorry for bothering you with this, instead of looking deeper into this myself. But I never used gdb before and it was quite some effort now, to make it work.
No problem. :)
So, bottom line, if anyone with a similar error ever reads this. Make sure that mpi4py uses the correct mpicc coming with PETSc. E.g.:
env MPICC=/path/to/mpicc pip install mpi4py
Thanks for the info!
r.
participants (2)
-
derj...＠web.de -
Robert Cimrman