Vaibhav Bora wrote: > Hey guys, > I am trying to run a 3-D simulation on a supercomputer. Each node has > about 2-4 GB of ram. > > When I run a small simulation it runs fine, but for the bigger > simulations, I get the an error... > > I think the error might be because we are running out of memory... Can I > specify the max memory / node that meep can use... >
As far as I know, there is no way to make meep use more memory in one computer than in another. I think it distributes the load evenly across the processors. Do you use dispersive materials? If so, you need to apply the following patch if you haven't done it already, otherwise the memory related to the dispersion will be stored on all the nodes in full instead of being distributed: http://ab-initio.mit.edu/pipermail/meep-discuss/2006-September/000376.html For the same problem, memory usage should be proportional to the number of points in the grid. Check how much memory it takes for a small problem, and extrapolate how much you would need for your real problem. Divide by the number of processors: that's the memory the cluster should have by CPU. Best regards, João Luís Silva _______________________________________________ meep-discuss mailing list [email protected] http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
