RE: Multithreading question

[Darold Higa]

Wow, thanks for all of the optimization tips.  I will definately give them a 
try.  Keep in mind that these "benchmarks" I'm talking about were for a very 
simple simulation that we used for that article some time ago.  It sounds like 
the data I was writing to the console for my co-author was the cuplrit.  There 
wasn't much of it, but it was probably enough to cause the performance hit he 
asked me about.  For my current simulation run I know a lot of my slowdown 
comes from I/O so I cannot avoid that.  The data collection for movies is 
pretty important for the kinds of simulations I am running.  Even when I turn 
hard disk access off and turn off 90% of the rasters, and even only display 
every 10th frame, it still takes a long time once my agent count gets high.

I started thinking about doing larger simulation runs for a paper based on my 
existing simulation in order to create a larger diveristy of competing agent 
strategies.  The problem is that with so many agents, even with very optimized 
code the simulation would take very long periods of time to execute.  In other 
words, I want to think about scalability.

One friend of mine mentioned distributed processing as a possible solution.  He 
volunteered to help me with this, but I wanted to come to the community first 
to see how big a hole I was digging for myself.  My first priority is to get 
some articles out so I can get a position somewhere, so that is why I'm sending 
out these feelers... :)

Darold Higa


                  ==================================
   Swarm-Support is for discussion of the technical details of the day
   to day usage of Swarm.  For list administration needs (esp.
   [un]subscribing), please send a message to <address@hidden>
   with "help" in the body of the message.