Cell splitting in neural networks extends strong scaling (Hines et al. 2008)

 Download zip file 
Help downloading and running models
Accession:97917
Neuron tree topology equations can be split into two subtrees and solved on different processors with no change in accuracy, stability, or computational effort; communication costs involve only sending and receiving two double precision values by each subtree at each time step. Application of the cell splitting method to two published network models exhibits good runtime scaling on twice as many processors as could be effectively used with whole-cell balancing.
Reference:
1 . Hines ML, Eichner H, Schürmann F (2008) Neuron splitting in compute-bound parallel network simulations enables runtime scaling with twice as many processors. J Comput Neurosci 25:203-10 [PubMed]
Model Information (Click on a link to find other models with that property)
Model Type: Realistic Network;
Brain Region(s)/Organism: Generic;
Cell Type(s):
Channel(s):
Gap Junctions:
Receptor(s):
Gene(s):
Transmitter(s):
Simulation Environment: NEURON;
Model Concept(s): Methods;
Implementer(s): Hines, Michael [Michael.Hines at Yale.edu];
/
splitcell
nrntraub
hoc
balcomp.hoc *
binfo.hoc *
defvar.hoc *
karkar.hoc
lbcreate.hoc *
loadbal.hoc *
mscreate.hoc
msdiv.hoc *
parlib.hoc
parlib2.hoc
traubcon.hoc *
traubcon_net.hoc *
                            
// The change to a connection coefficient gets changed back to
// its value determined by diam,L,Ra,topology after any change of any
// of those properties in any section.
// However the topology change implied by the traub_exact process is persistent.
// Thus one possibility is to do the traub_exact topology change
// along with the connection coefficient setting AFTER a complete setup
// that includes gaps, synapses, and stimuli, and then let NEURON do its
// thing in response to diam_changed,
// and then change all the connection coefficients.
// Another possiblity, which perhaps is not as efficient but is
// certainly simpler, is to
// let traub_exact do its thing on the creation of each cell, which will accomplish
// the persistent topology change, and save the info regarding the
// connection coefficients, and then fill them again after the complete setup.
// We choose the latter.

// for all cells
proc reset_connection_coefficients() {local i, gid, ix  localobj cell
	if (use_traubexact) {
		// do the topology first
		for pcitr(&i, &gid) {
			cell = pc.gid2cell(gid)
			ix = cell.type
			traubexact_topology(cell, traubExactInfo.tci[ix], traubExactInfo.traub_parent[ix])
		}
		doNotify()
		// now the coefficients
		for pcitr(&i, &gid) {
			cell = pc.gid2cell(gid)
			ix = cell.type
			traubexact_coef(cell, traubExactInfo.tci[ix], traubExactInfo.traub_parent[ix])
		}
	}
}

Loading data, please wait...