"Number of User Partition Cells" Error

General issues of interest both for network and
individual cell parallelization.

Moderator: hines

Post Reply
calvins
Posts: 6
Joined: Thu Jul 09, 2009 5:44 pm

"Number of User Partition Cells" Error

Post by calvins »

So I am new to NEURON and I have been running the parallel models in the ModelDB to learn parallel NEURON 7.0 on my Windows Vista OS (yeah I know...) before I attempt to learn it for UNIX. For the Migliore et al 2006 (Accession Number 64229), I have been trying to run the parallel version of the dentate gyrus model (Santhakumar et al), and if I even just open the Parallel Computing tool without changing anything, the following error message appears when I try to run the model:
nrniv: The total number of cells, 529, is different than the number of user partition cells, 0

in parRI10sp.hoc near line 1684
{doNotify()}
^
doNotify( )
xopen("parRI10sp.hoc" )
execute1("{xopen("parRI10sp.hoc")}" )
load_file("parRI10sp.hoc" )
and others
0
I was wondering if there was a fix for this error, because with other parallel models, such as Hines and Carnevale 2008 (96444), I have had no problem using the Parallel Computing tool. Thanks.

Calvin
hines
Site Admin
Posts: 1691
Joined: Wed May 18, 2005 3:32 pm

Re: "Number of User Partition Cells" Error

Post by hines »

The dentate gyrus, model is parallelized only for MPI. All of the mod files can be declared THREADSAFE
using
bash mkthreadsafe
except Gfluct2.mod which uses the normrand function.
Anyway, I am not experiencing the problem when starting the Parallel Computing tool with
the new alpha version at
http://www.neuron.yale.edu/ftp/neuron/versions/alpha/

I see that the use of Gfluct2 is commented out in the hoc files so one can declare Gfluct2.mod to be THREADSAFE
with mkthreadsafe (even though it is NOT) and one should be able to run this model with multiple threads and
obtain exactly the same result as with a single thread.
Post Reply