streeter@theory.lcs.mit.edu (Kenneth B Streeter) (04/08/91)
In article <24067@well.sf.ca.us> shiva@well.sf.ca.us (Kenneth Porter) writes: >I've received several inquiries about the Systems Concepts >CAM-6 coprocessor which I helped design many years ago. >talked to a friend who still works for SC and apparently SC is >in the process of moving to Nevada. SC has never considered >the CAM-6 to be a profit center so orders have been given a >very low priority (something I disagree with). While you can >still place an order, I would not expect to get any response on >it. At the time I left SC, I doubt that we shipped more than >30. Systems Concepts has essentially ceased shipping any CAM-6 boards. In response to this, a new company, Automatrix, has designed the "CAM-PC", a compatible redesign of the CAM-6. The CAM-PC will run experiments in CAM-FORTH for CAM-6, but also has some significant improvements. The complete CAM-PC package, consisting of the CAM-PC board, user manual, handware manual, cables, and the Toffoli & Margolus text "Cellular Automata Machines" is available from Automatrix for $1950. For further information on CAM-PC contact automatrix (518-877-7270, automtrx!campc@crdgw1.ge.com) or myself (addresses below.) >Some readers have asked me about a successor to the CAM-6 being >produced by another company and designed to run in a >SparcStation. This is the first that I have heard of this >project and would be interested in learning more. >I'm sure the new card will be much more sophisticated. Yes, the group at MIT that originally designed CAM-6 is now working on a successor to CAM-6. This successor is CAM-8, and is designed to interface to a Sun SPARCstation. A brief discussion of CAM-8 and its usage appears in Physica D 47 (1991) pg. 263-272: T. Toffoli and N. Margolus, "Programmable Matter: Concepts and Realization." If you cannot find a copy of this, I would send you a copy. The CAM-8 machine is highly configurable through software, naturally representing CA spaces in three dimensions, but configurable to anywhere from 1 to 16 dimensions. Each site in the CA space maintains 16 bits of state. The scan rate of the the basic CAM-8 module is 32 Mega-sites per second. Owing to the configurability of a CAM-8 module, the scan rate of the space depends linearly on the size of the space. Thus a 4 Mega-site space would be updated 8 times/second, or a 64K site space (256x256 for d=2) would be updated 500 times/second. The great benefit of the CAM-8 design is that multiple modules are intended to be connected, scanning the CA space in parallel. (Only one SPARCstation is required, regardless of the number of modules.) Thus, for example, when eight modules are connected together, 256 Mega-sites/second of update available, permitting a space 8 times larger to be updated at the same rate. If anybody is interested in further information on CAM-8, I would be happy to answer any questions you may have. -- Kenneth B. Streeter | ARPA: streeter@im.lcs.mit.edu MIT LCS, Room NE43-350 | UUCP: ...!uunet!im.lcs.mit.edu!streeter 545 Technology Square | (617) 253-2614 (work) Cambridge, MA 02139 | (617) 225-2249 (home)