NOBLE APE MAILOUT - FEBRUARY 2007
SIMULATED VERSUS ACTUAL
SIMULATED VERSUS ACTUAL
This month's mailout is going to be a little more abstract, primarily because there has been less hands-on development with the Simulation and more work relating to the surrounding components. Removing AdSense, working through the documentation for inclusion in the Simulation Manuals, inclusion on the site or removal - these things don't lend themselves to more than a singe mailout.
A regular theme through collective artificial life development - that's artificial life development that comes through talking with other artificial life developers and advanced users - is the idea of the supportive dichotomy between simulated and actual. One of these abstract ideas is that an artificial life developer can use some of the principles in their development to assist communicating their development. One of the characteristics of long-term artificial life development is an ability to tune a simulated system for sustainability. It doesn't take too much to tune the release cycle of an artificial life simulation for sustainability too.
These ideas relate to testing various factors and then implementing around these factors. This idea in moving from the theoretical to the applied is still up for debate. There is a great deal of idealism in the artificial life development. Artificial life developers tend to create environments that are self-justifying prophecies in some regards. The real world is not like that. Building on user numbers and growing user numbers organically is challenging. There are numerous factors that can not be categorized or analyzed until well after the fact.
INTEL AND DEMOSCENE
I was contacted this month by the folks at INTEL who are eager to use Noble Ape at Apple's WWDC 2007. The primary optimization is the brian simulation. This code has deviated quite dramatically from the continued Simulation development. The one-to-one testing for the brain shows that the vector code from Apple and INTEL probably never mirrored the scalar code in the brain simulation. With this one-to-one testing and the command line interface to the Simulation, I hope this year's WWDC demonstration will show true one-to-one representation of the scalar and vector code.
In addition the folks at INTEL very kindly passed on a version of their compiler. I had planned on purchasing the INTEL compiler sometime this year. I have discussed the expanding nature of compilers through the developer mailing list. The INTEL compiler's resultant Noble Ape Simulation indicated that I should review the source and reuse of source for the Simulation. Neither GCC nor the INTEL compiler could create a version of the Simulation under 100k.
Whilst this may still seem like a substantially small application I wonder what is stopping the reduction of size to something that is comparable to the CodeWarrior compilation or the Visual C++ compilation size. Reverting to my old instincts, I looked at the binary and found debugging string information in the final version. This can easily be stripped out but begs the question if reworking the Simulation to reduce compiled code size can produce further optimizations or if it is an exercise in futility.
Will Wright's failure to mention artificial life with the release of Spore is an ongoing topic of discussion amongst my artificial life developer cohorts. Wright has mentioned the Demoscene - a primarily Scandinavian movement who produce small applications that are non-interactive audio/visual presentations analogous to small digital performance art pieces.
The CodeWarrior compiled version of the Simulation could easy fit under the 64k requirements for the Demoscene. The interactive nature of the Simulation and the lack of audio make it an unlikely Demoscene candidate.
Hope all is well with you all,
Tom Barbalet, 28 February 2007.