Experience with SMC 8624T L2 switches and nics

Karen Shaeffer shaeffer at neuralscape.com
Thu Feb 27 09:59:35 PST 2003

On Thu, Feb 27, 2003 at 11:48:35AM +0100, Eugen Leitl wrote:
> On Thu, 27 Feb 2003, Felix Rauch wrote:
> > In the paper we are mainly interested in bandwidth (not latency) and
> > examine a few different networks on our 128 node cluster "Xibalba" [1]:
> An offtopic question. A friend of mine is writing spiking neuronal network 
> code, which are exquisitely sensitive to timing jitter and hate latency.

Indeed. One way to conduct such experimental simulations is to create a
virtualized, distributed time reference that is embedded in a distributed
application framework. This is a virtual simulation time and is completely
insensitive to real-time jitter and latency. Then your neuronal networks
triggering on differential delays are simulated based on this abstracted
time reference.

 Karen Shaeffer
 Neuralscape, Palo Alto, Ca. 94306
 shaeffer at neuralscape.com  http://www.neuralscape.com

More information about the Beowulf mailing list