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
pardentategyrus
readme.html *
bgka.mod *
CaBK.mod *
ccanl.mod *
Gfluct2.mod *
gskch.mod *
hyperde3.mod *
ichan2.mod *
LcaMig.mod *
nca.mod *
tca.mod *
bg.sh
DG500_M7.hoc *
dgnetactivity.jpg *
dgnettraces.jpg *
init.hoc
initorig.hoc *
modstat *
mosinit_orig.hoc *
out.std
parRI10sp.hoc
RI10sp.hoc
test1.sh *
time *
                            
setuptime = startsw()
if (name_declared("load_balance_phase") == 0) {
	execute("load_balance_phase=0")
}

{load_file("nrngui.hoc")}
{load_file("netparmpi.hoc")}
{load_file("../common/parlib.hoc")}

// use 0 for normal round robin,
// 1 to construct mcomplex.dat (or use mcomplex.hoc to construct)
// 2 to construct splitcell balance info file ("splitbal.dat")
// 3 for load balance using splitcell splitbal.nhost.dat file
// 4 whole cell binfo with cx.nhost.dat
load_balance(load_balance_phase)

ncell = 528
pnm.ncell = ncell
gid_distribute()

objref con_mat
con_mat = new Matrix(ncell, ncell, 2)

cells_count = 0

obfunc createcell() {
	cells_count += 1
	return par_create(cells_count-1, $s1)
}

proc nc_append() {
	con_mat.x[$1][$2] += 1
	par_ncappend($1, $2, $3, $4, $5)
}

func is_connected() {local c
        c = con_mat.getval($3+$4, $1+$2)
        return (c != 0)
}

{load_file("parRI10sp.hoc")}
nclist = pnm.nclist

{want_all_spikes()}
mkhist(50)

initNet()

{
cvode.cache_efficient(1)
pc.spike_compress(3, 1)
cvode.queue_mode(1,0)
}

setuptime = startsw() - setuptime

//if (pnm.nhost == 1) { saveNet() }


//tstop = 10
if (pnm.myid == 0) {print "SetupTime: ", setuptime}
if (name_declared("mosinit")) { stop }
prun()
if (pnm.myid == 0) {print "RunTime: ", runtime}
spike2file()
//if (pnm.nhost == 1) { SpkMx() }

{pnm.pc.runworker()}

print "Maximum integration interval: ", mindelay()
getstat()
prhist()
print_spike_stat_info()

{pnm.pc.done()}

perf2file()
quit()