On Thu, 7 Aug 1997 starling(a)umr.edu wrote: The document you mention below says that "only 350 machines were made (50 prototypes and 300 production machines or, according to documents from Atari, 100 prototypes and 250 production machines)." [Translation my own and may be inexact.] Definitely a rare beast! This sounds very much like Commodore's ill-fated transputer project. It also ran an OS called Helios, and used T800 processors. There's an article on Commodore's transputer project in AmigoTimes V1.2. AmigoTimes never liked putting dates on their magazines (dammit) but it's probably date mentioned in the article, and on the transputer board in the picture. Actually, the web article you directed us to states about the Atari machine that "it was first presented in November 1987 at COMDEX in Las Vegas under the name ABAQ", so it looks like the Atari machine was earlier. I've seen someone on Usenet say that they have some Commodore transputer boards, and I think he said that the project was cancelled after problems with the person or people who were writing the Helios operating system or something. Also the T-800 chips were never cheap enough to use in products for the kinds of markets Atari and Commodore generally looked for. I don't know how long ago it was that I saw that - it could still be on Deja News. A few quotes from the AmigoTimes article, "The Amiga Supercomputer" by Ernest N. Nagy: . . . "The operating system that will control the transputer chips, is called HELIOS. HELIOS is being written in 'C' by Perry Helion, specifically for use with the transputer and should be available by the fall of 1988." . . . "To comply with existing Amiga graphics standards, Xwindows will be used as the graphics interface. The Xwindow environment will be running through a shell command interpreter very similar to the UNIX shell." . . . "The advantage of a transputer outweighs its disadvantages. Each transputer has local RAM, the T-800 has an onboard floating point processor working at 1.2 MFlops and each chip (T-800) is rated at 10 mips. Each transputer communicates on an independent hardware level via four inter-processor communication links. Because transputers are self-contained and work in parallel, they have no bus bandwidth limitations, therefore 'n' transputers means 'n' times the performance. Unlike conventional CPUs linked together, transputers have linear growth potential. Each Amiga can be further expanded by up to four independent multitransputer cards. Each multitransputer card will contain four transputers (either the TMS414 or TMS800) with up to 4MB of on-board memory. These multitransputer cards will fit into the AT slot and an Amiga 2000 will be capable of supporting up to 17 transputers; providing a total of 170 mips in processing speed and 20 MFlops. The AT slots just provide the power to the board. An Amiga with a root transputer and one transputer board forms a parallel pipe configuration and two or more transputer boards allows two-dimensional movement capabilities from chip to chip. If 17 transputers is not enough power, then workstation superclusters can be implemented for very high performance application needs. To allow for full speed server code transfers, the workstation must be able to transfer data to and from the transputer in DMA mode. This will require a new device driver in the Amiga OS which will be optimized for HELIOS type data transfers. Presently link techniques are being developed that will allow data transfers between stations, at a rate of 20Mbit/s. There is a theoretical limit to the amount of transputers that can be controlled by HELIOS; it stands untested at about 30 workstations (or 500 transputers), but it may be more." . . . Then the article goes on to an interview with Dieter Priess, General Manager of Engineering at Braunschweig: . . . "AMIGOTIMES: We have heard about transputers before, but never with regards to the Amiga. The first we really heard of it, after INMOS brought it out, was after Jack Tramiel (Atari) announced that he is going to bring out a transputer based computer to compete with the success of the Amiga. When we got wind about a possible Amiga transputer expansion board, it was great news. PREISS: He is using the same operating system (HELIOS) as we are, so nobody is actually ahead of the timeframe, or behind. Except that we have some more plans with HELIOS than I have seen in his announcement at the moment. His machine has a closed architecture, and I think it contains 12 transputers. He thinks this is the kind of workstation that the future may need. We've found that limiting a computer to a given number of transputers is a violation of the transputer concept itself. Our policy has been, in the past, to keep machines as flexible and as open as possible. I really think this concept has to be applied to each expansion that we create for the Amiga. So both HELIOS, as well as the hardware, have to be open enough to support everything that is required by the demand of the user and not by our offer. AMIGOTIMES: Will the transputer board contain four transputers on the main board? PREISS: No, there are four transputers on the expansion boards that go into the PC slots; since we want to keep the Amiga slots free. The PC slots just supply the power for those boards, they do not use any bus signals. This means if you have four free PC slots in the system, you can add up to four expansion boards (16 transputers). Combined with the root transputer board in the Amiga slot, you end up with a total of 17 transputers." . . . "We personally saw the transputer root board working, so it's not 'vaporware', but HELIOS is not ready yet and until it is, the transputer board will continue to be improved. The entire system should be ready by the fall and hopefully be ready for commercial sale just before the Christmas shopping rush (the first week of December)." . . . It's a big article, and it's got a photo of the root board, and on it can be read the text: TRANSPUTER BOARD A2000 MADE IN W. GERMANY (BSW) (C)1988 COMMODORE Anyway, that's enough typing for tonight on such a vapourous (and not even quite classic) subject. :) Doug Spence ds_spenc(a)alcor.concordia.ca