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

 Download zip file   Auto-launch 
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]
Citations  Citation Browser
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
cells
hoc
mod
net
README
balcomp.hoc *
bgrunme
bgsmall.sh
bgsplit.sh
cell_templates.hoc *
clear.hoc *
finit.hoc *
fortmap.hoc *
gidcell.hoc
gidcell.ses *
init.hoc
manage_setup.hoc
metisbal.sh
mosinit_orig.hoc *
onecell.hoc
onecell.ses *
prcellstate.hoc *
prepare.sh
printcon.hoc *
spkplt.hoc *
vclampg.hoc *
vcompclamp.hoc *
vcompsim.hoc *
                            
load_file("stdlib.hoc")
if (!name_declared("pc")) { execute1("objref pc") }
if (!name_declared("pcitr")) {execute1("iterator pcitr() {}")}
strdef csf_name_, mtname_, msname_, varval_

proc prcellstate() {local i, gid   localobj c, csf
	if (numarg() > 1) if (!pc.gid_exists($2)) return
	sprint(csf_name_, "cs%d.%d.%d", $1, pc.id, pc.nhost)
	csf = new File()
	csf.wopen(csf_name_)
	for pcitr(&j, &gid, 1) {
		if (numarg() > 1) if (gid != $2) continue
		c = pc.gid2obj(gid)
		prcellstate_(c, csf, gid)
	}
	csf.close
}

proc prcellstate_() {local i, j, k, x, size localobj mt, ms, syn
	i = 0
	mt = new MechanismType(0)
	forsec $o1.all {
		for (x, 0) {
			$o2.printf("%d %d %.17g %s.v(%g)\n", $3, i, v(x), secname(), x)
			i += 1
			$o2.printf("%d %d %.17g area(%g)\n", $3, i, area(x), x)
			i += 1
			$o2.printf("%d %d %.17g ri(%g)\n", $3, i, ri(x), x)
			i += 1
		}
		for j=1, mt.count-1 {
			mt.select(j)
		mt.selected(mtname_)
			if (ismembrane(mtname_)) {
				ms = new MechanismStandard(mtname_,0)
				for k=0, ms.count-1 {
					size = ms.name(msname_, k)
				for (x,0) {
sprint(varval_, "hoc_ac_ = %s(%g)", msname_, x)
					execute(varval_)
$o2.printf("%d %d %.17g %s(%g)\n", $3, i, hoc_ac_, msname_, x)
						i += 1
					}
				}
			}
		}
	}
	for j=0, $o1.synlist.count-1 {
		syn = $o1.synlist.object(j)
		classname(syn, mtname_)
		ms = new MechanismStandard(mtname_,0)
		ms.in(syn)
		for k=0, ms.count-1 {
			ms.name(msname_, k)
$o2.printf("%d %d %.17g %s.%s\n", $3, i, ms.get(msname_), syn, msname_)
			i += 1
		}
	}
}

proc rdcellstate() {local i, g, j, val1, val2, diff, most localobj f1, f2, l1, l2, lmost
	most = 0
	lmost = new String()
	l1 = new String() l2 = new String()
	f1 = new File() f1.ropen($s1)
	f2 = new File() f2.ropen($s2)
	for (i=0; !f1.eof; i += 1) {
		f1.gets(l1.s)  f2.gets(l2.s)
		sscanf(l1.s, "%d %d %lf", &g, &j, &val1)
		sscanf(l2.s, "%d %d %lf", &g, &j, &val2)
		if (val1 != val2) {
			diff = abs(val1-val2)/(abs(val1) + abs(val2))
			printf("%g %s\n", diff, l1.s)
			if (most < diff) {
				most = diff
				lmost.s = l1.s	
			}
		}
	}
	if (most > 0) {
		printf("most %g %s\n", most, lmost.s)
	}
	f1.close
	f2.close
}