CAM-8 continues to be the rallying point for our CA research. The object of the CAM-8 hardware project was to produce machines that are well suited to spatially organized, local, non-numerical calculations such as lattice-gas simulations of physical systems. We took a virtual-fine-grained approach, with emphasis on scalability, cost/performance, and visualization/analysis capabilities. The result was that in our small-scale prototypes, we take an amount and kind of hardware comparable to what you might find in a high-end personal computer, and arrange it into a personal CA supercomputer. These small prototypes run typical lattice-gas simulations at a speed and on a scale comparable to the best simulations that have been done on any machine.
Our small prototypes have been running for about 2 years now, and we have several of them out in the field (with collaborators who've paid for them), with another half dozen slated to go out to various research groups in the U.S., Canada, Europe, and Asia in the next few weeks (including one to the Media Lab).