Re: [Swarm-Modelling] SWARM on Clusters

[Marcus G. Daniels]

Russell Standish wrote:

> The idea behind Graphcode is "agents on a graph". Graphcode takes care
> of migrating agents between processors as necessary, including for
> load balancing reasons, and performing the work on a local list of
> nodes. Nearest neighbours living on distant nodes are cached prior to
> running the computation part.
>
>  
Cool, sounds kind of like Tom Maxwell's SME package.  [Don't know if 
there is a maintained version these days..]
I've been playing with a package called POOMA lately.  
http://www.codesourcery.com/pooma/pooma
It was developed at Los Alamos National Labs, and implements data 
parallel arrays & fields.  It lets you write numerical simulations in 
high level code (hidden behind operator overloading in C++), and then it 
let's you finesse the partioning of the computation over the cluster.

> One of the things we did figure out how to do was to embed C++ models into Swarm. 
Swarm implements callin and callouts from XPCOM just like the Java can 
do callins and callouts.  The nice thing about this system is that the 
XPCOM interfaces are C++ classes and Swarm can call these methods at 
function call speed.  The XPCOM support also facilitates writing Swarm 
models in JavaScript and Python.  It's pretty much complete -- probe 
stuff is implemented for XPCOM and JavaScript, Selectors work 
(reflection), etc.  If I were to dust it off again I'd probably target 
Microsoft COM or .NET since the technology is 90% the same.   It would 
be like Swarm for Java, but instead of plugging Swarm into your Java 
IDE, you'd plugin it into Visual Studio and have Swarm interfaces in 
Visual Basic, Visual C++, C#, etc.