Re: Using Breve, Digital Space, Framsticks et al

Quoting Gerald de Jong <[email protected]>:

>> I've been talking about a DLL version of Noble Ape for about... two years
>> now... for some interface to Digital Space. In a similar timeframe talking
>> with Jon about Breve.
>
> what's the showstopper?

There are a number of parts to Noble Ape;

* the external environment simulation - this creates the landscape and  
the weather.

* the ape movement - this can be controlled by the hardcoded  
simulation or through;

* ApeScript,

* the cognitive simulation - which can be switched on or off depending  
on how you want to use it with the ape movement.

* the platform independent graphics - non-polygonal, lines,  
characters, meters etc.

The trivial implementation of Noble Ape in BREVE would take the ape  
movement and rewrite it in Steve. However I'm far more interested in  
including at least the movement and the external environment  
simulation. In order to do this, I would need to implement both these  
things in C/C++ with an interface layer into Steve. This is doable,  
but finding the right kind of interface was the technical challenge.

Digital Space I'd like the external environment + aper movement but  
most importantly;

* forms - polygonal Noble Apes (which aren't currently part of the  
Simulation but coming soon!)

>> So the good news is the hyper-threaded over-network version is nearly ready
>> for release. This still brings into question whether that version will be on
>> the DLL/library or whether it will be the old version etc.
>
> can you explain how hyperthreading and network communcation are
> playing a role in your simulation?

So in terms of current processor usage;

20% - the external environment simulation (mostly the weather)

tiny - the ape movement

30% - the cognitive simulation - which can be switched on or off  
depending on how you want to use it with the ape movement.

30% - the platform independent graphics - non-polygonal, lines,  
characters, meters etc.

remainder - OS, mouse interactions etc.

It turns out the external environment can be completely deterministic  
- no random factors. The only random factors are tracked per ape. Each  
ape can be farmed out over a network if its "randomiser" is tracked  
locally and it communicates per time cycle (or so) if anything happens  
of interest.

INTEL basically looked at these processor-use numbers and said - farm  
out the heavy work to other processors, over a network and to a  
graphics processor. QED - instant silicon nirvana.

Best regards,

Tom Barbalet.


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/