Parallel sequences should work like a charm: for example MASON's
HeatBugs example uses one.
We recently had some discussion on the list about making sure that the
parallel sequence's threads are properly gathered by simulation's end,
but that shouldn't affect you at all.
The entire of a parallel sequence will be executed in one step. So
for example if you want all your fish to gather read-only (be
threadsafe remember) information about their world, and have four CPUs
on your machine, you could create a ParallelSequence which holds four
Sequences, and load your fish, roughly balanced, into those four
Sequences. Each time the schedule steps, all the fish will be stepped
once.
Some possible places where you might be making an error: any chance
you're rescheduling your fish agents rather than rescheduling the
ParallelSequence? And don't use a RandomSequence inside a
ParallelSequence.
Sean
On Jul 19, 2010, at 3:31 PM, Steven Saul wrote:
> Hi Sean,
>
> Quick question: should a parallel sequence work when running my
> simulation using the GUI? I use the GUI just to de-bug and make
> sure I
> coded something correctly. With the parallel sequence, the simulation
> seems to go from start right to the end without stepping anything. It
> runs fine on the command line. I assume this is a coding error on my
> side, but just wanted to make sure.
> Otherwise, I nearly have the whole sim parallelized including events
> that
> don't happen each time step such as initialziation of objects and
> recuitment events.
>
> Thanks,
> Steve
>
>
> --
> Steven Saul, M.A.
> Graduate Assistant, Marine Biology and Fisheries
> Cooperative Institute for Marine and Atmospheric Studies
> Cooperative Unit for Fisheries Education and Research
> University of Miami - RSMAS
> 4600 Rickenbacker Cswy.
> Miami, Florida 33149
> + 1 305-421-4831
> http://cufer.rsmas.miami.edu
|