This architecture, inspired by the work of both CarverMead and KungAndLeiserson, solved many addressing, indexing and syncronization problems by simply unwrapping the algorithm over the surface of the chip.
In the early 80s I wrote a simulator as an object of study in an advanced Smalltalk class. One of the students was a researcher from Dow Chemical. Although he came to learn Smalltalk, he took away a committment to build a process control envirnment that was as simple as my simulator. I met him several years later at a conference where he told me that he had reduced my simulation to a GaAs microcontroller that, in addition to controlling a chemical process, could communicate with other controllers over optical fiber and generate a video visualization of its current state.
Inspired by TomRay, I wrote a simulator to see if I could get programs to evolve. I organized memory as characters on a plane and offered, again, adjacency as the only addressing mode. Although it was fun to program, I found two dimensional programs hard to copy.IanOsgood has translated this work to javacript.
I was taken by the power and simplicity of apple's CocoaWorld research. This simulation environment continuously applied transformation rules to the elements in a two dimensional space. I've written up my implementation of Knuth's elevator simulation as a LittleSimulatorInCocoa. I've also computed FibonacciNumbers with selfish digits and modeled chemical and cytoskeletal reaction dynamics in Cocoa.
In the spring of 2002 I attened (and co-hosted) the Biological Framings of Problems in Computing workshop at the Santa Fe Institute. In prepairation I wrote a manifesto called EscapingAddresses which, along with an excellent lecture on SignalTransduction by WalterFontana, form the inspirational foundations of the Cybords project.
Some work by others has caught our attention.
|Last edited April 11, 2008
Return to WelcomeVisitors