Re: Parallel processing

[Tony]

Glen, 

Appreciate the response.  Perhaps I can respond with a diatribe of my own
<grin>  to light the shadows.

When I referred to Linux, I really meant Linux, but I did understand that
it is Unix on which you are concentrating.  My inexperience with parallel
processing, as well as not fully understanding the differences between
Linux and Unix cloud my understanding.  I'm all for portability and don't
think Swarm should veer off to a particular OS, whether it's NT or Linux or
OSBloat x.x  The question I really should have asked (for anyone to
respond):  what is the feasibility of parallel processing on multiple
pentiums running Linux and Swarm?  How parallel friendly is Swarm
currently?
What is driving these questions is musing about how much horsepower it's
going to take to run simulations with bunches (>100 or even >1000) of
agents that have multiple hierarchical swarms themselves.  Anyone have any
benchmarks or feedback on this?

Now here's where I go off the deep end :P

If you'll look back at my two previous postings, you'll note that I am
sensitive to the administrative overhead of posting project information and
that the Swarm team has less than 2 full time people working on it. 
*However* <grin>, I am astounded that a formal project outline / task
breakout is not yet available.  I know that you've recently solicited
surveys from the Swarm community asking for input, but wonder why there
isn't a plan for the next 6-12 months.  Who has the master plan?  What is
the decision making process between CGL and the rest of the team?  What are
the weights of community inputs?  Does NRL / ARPA have a voice or a
Statement of Work to which you have to stick?  Is Elvis in orbit with a
race of space vampires?  I think it's well understood that the long range
goals will change and that some minor tasks will fluctuate wildly, but we
really need this project structure.

My biggest heartache with the internet and the web is that we often don't
realize the administrative overhead required to keep communities of
interest appraised of what we're doing, whether's it's a commercial site or
a software development site.  My other observation is that information
posted on the web begets a hunger for even more information.  Even
well-developed and consistenly maintained sites are often frustrating in
omissions of information. 

Here endeth the rambling,
Tony