Name that Ware, March 2015

The Ware for March 2015 is shown below.

Thanks to Dale Grover for sharing this ware! I had read about this one as a lad, but never laid hands on one…

21 Responses to “Name that Ware, March 2015”

  1. tz says:

    Vesa Local Bus (or some other ISA derivative) video card? RGB output (sync on green), maybe something like a 8514, with the lightpen at top, half populated memory?

    Otherwise some kind of network card, but the connectors are wrong. I only say it because of the “link speed” selection. But it isn’t PC-NET or Token Ring. And I don’t see the right stuff for thinwire ethernet.

  2. melstav says:

    Damn, that was quick.

  3. J. Peterson says:

    I was also going to say early-90s video card, but the Transputer guys are right. A video card would have a bit more analog near the RGB connectors, and by then you’d see specialized chips (e.g., Brooktree RAMDACs) in front of the analog output.

    So instead, the connectors on the right are for networking the Transputer cards.

  4. f4eru says:

    This is a Perkin-Elmer INMOS Transputer Board

    a newer model :

  5. David Given says:

    It’s broken, that’s what it is — that chip bottom right needs reseating…

    I programmed for a transputer a few times; firstly, learning Occam at university, which was awful (the entire IDE ran on a single node, which was way too slow to cope). Then later at Tao Systems, where we had a asymmetric multiprocessing OS which would run on a nine-node system (eight transputers and a Pentium — the PC was so much faster than the transputers that when running our standard demo, a parallel mandelbrot generator, the Pentium could render 632 scanlines in the time it took the transputers to render their single scanline). And then finally, after Transputer died and their processor was reinvented as the ST20 embedded microcontroller. We were compiling C on it. It was… quite special.

    • David Given says:

      …also I forgot to mention my favourite transputer war story: at Tao we discovered that if we tried to shift an integer right by 0 bits, the whole thing would hang with interrupts off.

      On studying the microcode we discovered that the right shift instruction was implemented as, more or less:

      do {
      register >>= 1;
      } while (shift);

      We calculated that the processor would probably have woken up again after about a fortnight.

      • Roger Shepherd says:

        It certainly the case that shift by zero places had this effect in the prototypes but I thought this was fixed in the production transputers. It certainly was the case that the implementation was bit-per-cycle and there wasn’t a check for long shifts. But the figure of a fortnight is wrong – at 20 MHz it takes about 3 minutes to shift 2^32 places.

        Now there was one bug which could cause the timer to hang for 2^32 microseconds (a bit over an hour), that was fixed.

        Actually, at 20MHz it would have been a little over 3 minutes as it shifted a bit per cycle.

        • Roger Shepherd says:

          I meant to add an explanation. The reason for not doing the checks for long shifts was that it would have made shifts take an extra cycle and the check isn’t needed in the common case of shift-by-a-constant. Correctness (which depends on your definition of the correct behaviour of long shifts) is achieved by a short code sequence. Now, that doesn’t mean that instructions which are capable of DOS are OK, with hindsight that behaviour was a mistake.

    • Felix says:

      I’ve worked with an ST20 compiler that, and I’m not kidding nor exaggerating, was _fixing_ typos heuristically.

      When the compiler couldn’t find a symbol, it would attempt to find a similar symbol (not just different case, but also longer/shorter variants) and _only emit a warning_ that a slightly differently named symbol was chosen.

      Worse, the project I’ve (eventually decided to not) work on was _relying_ on this feature to even compile.

      Horrible code.

  6. mb says:

    I knew I used one for connecting instruments way back when in an environmental lab, oh well maybe next time I will remember more quickly

  7. gamemanj says:

    As much as all the Transputer guesses are nice, I’m just going to go with a guess, because really if I just said “Transputer” that would be unoriginal.
    I’m going with a old ISA networking device of some kind, 3 ports because either the protocol needs 3 shielded lines(sync/in/out) to work or because they’re 3 1-directional links for goodness knows what purpose.

  8. josh says:

    It’s not a right place to ask, but the biology category too long to leave a message. My question is that: is there any books or tutorials for computer engineer and bios dummy to make a DNA analysis device?

  9. we provide rugged smart terminal, include rugged tablet, rugged handheld terminal. multifunction include RFID UHF 13.56MHz, Barcode reader, Finger printer, infrared ray and so on. we also use Gorrila, TFT screen on it. every function and size can customize

  10. Andy says:

    I’m interested in testing the transputer board in my PE Spectrum 2000 FTIR. Would it be possible for someone to provide me with one to use?

  11. That’s a composite TARGA frame buffer. Did a wonderful 512×512 NTSC picture. (I know the NTTSC actual image is 480 high). It was from the early days of video. Cool sidenote, I used them on the Grateful Dead’s last tour with Jerry Garcia. Those boards dis projector masking. Not a real time board, you rendered a frame and then had a VTR record the frame, rerender, advance the tape a frame and record. Rinse Repeat.