Official SWITCH-ON DAY of CAM-Brain Machine (CBM)
29 November 1999
After a year or so of delay due to various technical and commercial reasons (e.g. Xilinx taking off the market the XC6264 chips that the CBM is based on - having to test these chips that we were able to extract from Xilinx - fabrication faults with ball grid array chips etc), the CAM-Brain Machine (CBM) finally had its switch on ceremony on the 29th November 1999.
The CBM really does update cellular automata cells at a rate of 130 billion a second. It really does evolve a neural net circuit module of some 1000 neurons in a second or so. It will really be able to update a 64000 module artificial brain in real time. Hopefully now, a new era in neural networks, artificial life, evolvable hardware, etc will begin. Even better, a new research field and industry will grow up called "brain building", based on this fabulous new machine.
Now that the machine is running, the learning curve of the Brain Architects (BAs) and the Evolutionary Engineers (EEs) using it can begin. We will need to see how well it evolves various circuits and how well it allows multimodule systems to function. Now that the machine is finally operating, progress should be rapid. Many journal articles on its performance and what it can do should appear in the year 2000.
There are only enough XC6264 chips to build 6-7 CBMs worldwide. As of end November 1999, 3 to 4 machines have been allocated. I would like to reserve one further machine for the US and another for China. Then there are no more until the second generation CBM (CBM-2) is completed in 3-4 years. If the Americans or Chinese want a machine, they had better buy one pretty fast.
Prof. Hugo de Garis and Dr. Michael Korkin with the CBM on "Switch-On Day" (29th Nov 1999)
The switch-on of the CBM attracted a lot of attention
Some of the 72 electronic boards in the CBM, each holding a Xilinx XC6264 chip
de Garis and Korkin with their CAM-Brain Machine (CBM), 1999 - a historic photo one day?