Problem updating connections with structural plasticity and multiple threads

[sdiazpier]

Unwanted behaviour is observed when updating the connectivity of the network using structural plasticity with multiple threads. This is not the case when multiple MPI processes are used. It seems like it is a problem with the structural plasticity manager while updating the number of deleted synapses.

[heplesser]

@sdiazpier Could you add a script reproducing this problem? It should be as compact as possible and preferably written in SLI so we can develop a regression test from it.

[heplesser]

@sdiazpier What is the status on this?

[heplesser]

Per 2.12, NEST should at least prohibit simulations with structural plasticity and multiple threads, if we cannot solve this issue properly by then.

[sanjayankur31]

@sdiazpier - can you please expand on "unwanted behaviour"? I'm using SP with MPI but I haven't really been analysing the data yet. If you can give us more information, I can narrow down the issue. Either a test case or a data dump from a test case would be most helpful?

[heplesser]

@sdiazpier Could you create a follow-up issue to remind us to check whether and how structural plasticity can be enabled in multi-threaded simulations?

[heplesser]

Using structural plasticity is now blocked when using multiple threads (see #629) and #633 is created as a follow-up issue. Thus closing this issue.