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]
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 *
                            
objref vmat, fvmat, voutlist, vinlist, fvoutlist, tout, tin
proc vcompclamp() { local i, j, numcomp  localobj vv, f, s, clist, c, rsav, cdvdt
	s = new String()
	voutlist = new List()
	vinlist = new List()
	fvoutlist = new List()
	clist = new List()

	f = new File()
	numcomp=0 forsec cell.all numcomp += 1
	vmat = new Matrix(100*100-1, numcomp+1)
	classname(cell, s.s)
	sprint(s.s, "../p2c/state/v_%s.dat", s.s)
	f.ropen(s.s)
	vmat.scanf(f, vmat.nrow, vmat.ncol)

	fvmat = new Matrix(100*100-1, numcomp+1)
	classname(cell, s.s)
	sprint(s.s, "../p2c/state/fv_%s.dat", s.s)
	f.ropen(s.s)
	fvmat.scanf(f, vmat.nrow, vmat.ncol)

	i = 1
	tout = new Vector()
	cell.comp[1] {tout.record(&t)}
	tin = vmat.getcol(0)
	rsav = new Vector()
	for i=1, vmat.ncol-1 cell.comp[i] {
		rsav.append(Ra)
		Ra = 1e9
		c = new SEClamp(.5)
		c.dur1 = 1e9
		c.rs = .01
		clist.append(c)

		vv = new Vector()
		vv.record(&c.i)
		fvoutlist.append(vv)

		vv = new Vector()
		vv.record(&v(.5))
		voutlist.append(vv)

		vv = vmat.getcol(i)
		vv.play(&c.amp1, tin, 1)
		vinlist.append(vv)
	}
	stdinit()
	continuerun(100)
	for i=1, vmat.ncol-1 cell.comp[i] {
		cdvdt = voutlist.object(i-1).c.deriv(dt,2)
		fvoutlist.object(i-1).div(cm*area(.5)/100).add(cdvdt)
	}
	for i=1, vmat.ncol-1 cell.comp[i] {
		Ra = rsav.x[i-1] // but no longer is traub_exact
	}
	seefv(2)
}
proc seefv() {localobj s
   s = new String()
   gg.erase_all()
   seewhich = $1
   if (seewhich > fvoutlist.count) {seewhich = fvoutlist.count}
   if (seewhich < 1) { seewhich = 1 }
   cell.comp[seewhich] { sprint(s.s,"%s(.5) clamp current", secname()) }
   gg.label(.5,.8,s.s,2,1,0,0,1)
   fvmat.getcol(seewhich).line(gg, tin, 2, 1)
   fvoutlist.object(seewhich-1).line(gg, tout) 
}	

proc mkseefv() {
	xpanel("compare compartment currents")
	xvalue("which", "seewhich", 1, "seefv(seewhich)")
	xpanel()
}




Loading data, please wait...