Re: [Swarm-Modelling] ABMs on Graphical Processor Units

[docgca]

I see that this was presented at Agent 2007; was there much cross pollenation between that group of attendants and those that would be expected to come to Swarmfest?  I'm just thinking (in terms of the program for Swarm 2008) that there are a lot of interesting issues regarding the methodological aspects of ABM (I'm also thinking of the recent thread between Miles and Marcus re: metaABM) that might warrant a designated portion of a day for these topics.  I wanted to float that among the general community; maybe the board can have an offline discussion in more detail as well.

Gary An

-----Original Message-----
From: Miles T. Parker <address@hidden>
To: Agent-based modeling <address@hidden>
Sent: Sun, 9 Dec 2007 12:34 pm
Subject: Re: [Swarm-Modelling] ABMs on Graphical Processor Units

Yeah, this is great! This is something a number of folks in the community have discussed -- going back to an early Swarmfest where the target was arrays of PS1s. But of course, this somehow never progressed beyond the level of "bar table engineering". :) Nice to see someone finally implement it and very elegantly. 
 
On Dec 9, 2007, at 10:17 AM, Marcus G. Daniels wrote: 
 
> Steve Railsback wrote: 
>> "our system can run 4 orders of magnitude faster than Ecolab on >> large versions (2 million + agents) of test case ABMs" 
>> 
> Awesome. Now if we just had Swarm running on Cell processors.. 
> _______________________________________________ 
> Modelling mailing list 
> address@hidden 
> http://www.swarm.org/mailman/listinfo/modelling 
 
_______________________________________________ 
Modelling mailing list 
address@hidden 
http://www.swarm.org/mailman/listinfo/modelling 

---

More new features than ever. Check out the new AOL Mail!