Closed Laamkh closed 4 years ago
This may sound trivial but are you sure that the path to the executable ~/athena/bin/athena is correct, and this is properly configured with MPI? It looks like you used an executable in a directory named "18".
This may also sound trivial, but did you perform "make clean" after reconfiguring the code? The code can fail if object files from previous configuration exist.
It seems mpirun recognize the whole line include the arguments as the executable due to the backslashes. Maybe try getting rid of the backslashes?
This may sound trivial but are you sure that the path to the executable ~/athena/bin/athena is correct, and this is properly configured with MPI? It looks like you used an executable in a directory named "18".
This may also sound trivial, but did you perform "make clean" after reconfiguring the code? The code can fail if object files from previous configuration exist.
Sorry for my late reply. The compilation should be fine and I am sure I have performed make clean. I put the Athena executable in `/athena/bin/athena while the 'athinput.kh-shear-lecoanet' in ~/work/kh/18.
Now, I try to put them in the same directory and delete the backslashes. It works this time.
Thank you so much for your suggestion.
Sorry, this may be a stupid question to ask here. I don't know if someone here can help me or I should consult someone who are familiar with the system of my computer.
I was trying to run the Kelvin-Helmholtz problem with Open-MPI. The problem generator function in the
kh.cpp
file was modified. It was the first time for me to use Open-MPI, I guess I have done something wrongly so the segmentation fault appear even when I configure Athena++ without the-mpi
flag but there was no such problems before the first trial of using MPI.Below is what I have done and the output
As I failed to run with MPI, I tried
Ever since, even when I configure the code without the -mpi flag, segfault appear. Is that I have altered the environment of my computer?
Thank you for your effort.
Regards, Laam