Closed DMurataj01 closed 1 year ago
Considering the patch partitioned also when running on a single MPI process was done on purpose. This allows to test MPI code in the very simple case of just one process. I am open to change this, however can you share some details on the segmentation fault you are seeing when running on a single process?
Considering the patch partitioned also when running on a single MPI process was done on purpose. This allows to test MPI code in the very simple case of just one process. I am open to change this, however can you share some details on the segmentation fault you are seeing when running on a single process?
Would love to help you but this particular issue was seen by a colleague. I think it makes sense to be able to set n procs to 1 when benchmarking and it follows a serial routine. - @andrea-iob maybe have a debugging override ?
This change has been merged through #365.
This fixes a segfault when attempting to run with one MPI rank