Re: The Swarm space (was RePast (another agent toolkit) released)

[Marcus G. Daniels]

>>>>> "MP" == Miles Parker <address@hidden> writes:

MP> This frankly is me as proud father putting Ascape in the
MP> best possible light - - the sugarscape model is an extreme example
MP> that takes advantage of a number of Ascape optimizations. 

>From one perspective, it's good if the various ABM toolkits focus on 
performance in a practical way, i.e. if anyone should be fast at
SugarScape, it ought to be AScape, eh?

However, it would be a pity if we start bending designs around naïve
performance questions and comparisions.  The only good I can see
coming from that kind of effort is that makes users more aware of
common bottlenecks, and their associated infrastructure: "Boy, if you
want fast permuted Java callouts over homogenous sets, use Swarm's
FActionForEachHomogenous interface".  It is probably possible to use
heatbugs or SugarScape or whatever as a benchmark, but people should
realize that it says as much about the coding skill that goes into
making the model efficient as it does the underlying infrastructure.
Pedagogically useful and fast models are not necessarily going to be
written in at all the same way.

MP> And prob. if were were running on a Linux box !  with an open
MP> source vm, and Swarm wasn't running under the Cygnus layer, Swarm
MP> would do better.

Cygwin is slow for system calls, but otherwise it doesn't extract any
significant cost across the board.  My experience is that batch
simulations using Cygwin are among the fastest (across platforms)
since Windows DLLs don't have the overhead of position independent
code.

MP> For the most part, these runs are pretty graphics bound, and since
MP> Java's weak point to some extent is still graphics, this is
MP> prob. where any perforance hit comes in. Computationally, I
MP> haven't done any real comparisons. My models run _much_ faster
MP> (typically by multiple orders of magnitude) when the (run-time
MP> pluggable) graphical views are removed, but I'm sure the same is
MP> true of the others.

..and it's reasonable to pay for performance with portability, especially
with the graphics aren't crucial to the modeling.

MP> While performance is an issue here, it is not as important IMNSHO as
MP> other things..and in any case needs to be looked at in the context
MP> of other hardware and software issues.

Amen!


                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================