Dear all,
As I understood from previous posts on the google groups, we can use the "parallel" facet of the experiment to run multiple simulations in parallel. But if I just wanted to run the one simulation, what are the scalability options available? **(All the discussion is pertaining to the headless mode)**
Does a single simulation use all the cores on the machine? The headless mode output does start my listing the number of available CPUs, but after reading the
runtime concepts it is not clear if a single simulation can and does utilize multiple cores.
Further, can a simulation be scaled to multiple nodes like one does using MPI?
My simulation takes 3 hours with 1200 EV agents, and around 6 hours with 2000 EV agents. So the paper reviewers have raised concerns about scalability. While, I am working on optimizing the runtime using better GAMA constructs, I think also exploring the distributed computation possibility feels prudent.
Thanks