Re: RE: NONMEM/PsN benchmark for SGE expansion

On 4/1/2011 8:33 AM, Jurgen Bulitta wrote: > b) If you have 2000 nodes, it may be worth to consider launching jobs > from several > master nodes. You can run into trouble both with the available RAM and > with the > network traffic, if everything needs to funnel through one master node, > even if you > use 4x Infiniband networking, for example. I think this is primarily a consideration that needs to be done in a cluster (based on MOSIX) rather than on a grid (based on OGE/SGE). On MOSIX, every process is started on the/a master node and then migrates away to the "best" (in terms of resources) computing node. On OGE, job specifications are stored in a queue, and started directly on the best computing node. Of course, the master in an OGE grid should be reliable (consider one or several shadow masters), but to my understanding, RAM and network bandwidth are less of a bottleneck on OGE than on MOSIX. (For that, MOSIX has other advantages...) Kind regards, -- Paul Matthias Diderichsen, PhD Quantitative Solutions B.V. +31 624 330 706