|Department of Computer Science||University of San Francisco|
Parallel NeuroSys is a collection of programs for the simulation of large, biologically accurate, neuronal networks on parallel computers and networks of workstations. It was originally developed by the University of San Francisco's Applied Mathematics Research Laboratory. Professors Peter Pacheco and Marcelo Camperi of USF are continuing to develop the program.
In order to use the programs, you will need an implementation of MPI. The MPICH implementation can be downloaded from Argonne National Lab, and the LAM implementation can be downloaded from Notre Dame University. The visualization program, neurondiz, currently requires that you have X-Windows installed and Argonne's MPICH implementation.
Further information on Parallel NeuroSys is available in the paper "PARALLEL NEUROSYS: A system for the simulation of very large networks of biologically accurate neurons on parallel computers" by Peter Pacheco, Marcelo Camperi, and Toshi Uchino. This paper is available in compressed postscript. A slightly revised version of this paper appeared in the June 2000 issue of Neurocomputing. This version is available in PDF.
Please note that since the paper appeared, we found a bug in the differential equation solver. Correcting the bug only affected reported runtimes and parallel efficiencies. In general, runtimes are improved. However parallel runtimes aren't proportionately improved. So parallel efficiencies have been reduced.
You can download the software by clicking
here. (The bug in nscanf
has been corrected.)