I would like to know if it is possible to speed up a simulation using parallel threads.
We have used before:
param name="numberOfThreads" value="96"/>
and this helped us to speed up the simulation on a single node.
Is there any parameter that we can change to make use of multiple nodes, or in general to make the simulation faster?
Hi Michele, I doubt you will get much improvement after 16 threads. I am not so sure about multi-node simulation, but I know Pieter Fourie completed a pseudosimulation (best to read about that in the MATSim book). Also, simplifying the network (first just 'clean' it) or running smaller samples, say 10%.