Dear Michele,
I assume you use the hpc_benchmark.py without any modifications? On my laptop,
mpirun -np 2 python install/share/doc/nest/examples/pynest/hpc_benchmark.py
executes without problems with NEST 3.6 (and with current master).
Interestingly, for the hpc_benchmark, NEST should never even get to the else block to which the assertion on line 107 in target_table.cpp belongs, since all connections
in the network are primary connections. So something seems to go wrong in the communication of information about connections to the presynaptic side.
Could you try with this branch
https://github.com/heplesser/nest-simulator/tree/36_nosingle
It makes sure all MPI communication strictly happens in the OpenMP master thread (in NEST 3.6, it may happen inside OpenMP single constructs). This should not make a
difference for the hpc_benchmark, since it uses a single thread by default.
Best,
Hans Ekkehard
--
Prof. Dr. Hans Ekkehard Plesser
Department of Data Science
Faculty of Science and Technology
Norwegian University of Life Sciences
PO Box 5003, 1432 Aas, Norway
Phone +47 6723 1560
Email hans.ekkehard.plesser@nmbu.no
Home http://arken.nmbu.no/~plesser