• MAP - Manufacturing Automation Protocol (Re: The Weave)

    From Lars Poulsen@3:633/10 to All on Sun Aug 9 15:39:06 2026
    On 2026-08-09 13:29, Lawrence D?Oliveiro wrote:
    On 9 Aug 2026 09:39:19 GMT, Stefan Ram regurgitates more AI slop:

    Capable of chaining up to 15 instruments simultaneously over
    rugged, shielded cables with substantial parallel throughput, it
    bypasses the unpredictable latencies and CPU-heavy software
    overhead that plague modern USB and network stacks.

    Back in the early 1980s, big companies like Boeing and GM proposed
    protocols called ?MAP? and ?TOP? to handle real-time communication on
    the factory floor. They didn?t trust Ethernet, you see, because it
    didn?t seem like it could guarantee reasonable bounds on latency,
    throughput etc.

    People still laughed at collision-based protocols then. Who?s laughing
    now?

    <https://en.wikipedia.org/wiki/Manufacturing_Automation_Protocol>

    Back in those days, I worked at ACC. While we made all our money from
    systems and devices to support BBN-1822, ARPAnet, MILnet, DISnet etc,
    our commercial marketing people insisted that the future belonged to the
    OSI protocol (Open Systems Interconnection) family. MAP/TOP was a
    framework to build engineering and manufacturing systems on that
    foundation. We spent hundreds of thousands of dollars to produce
    prototypes of "intelligent front-ends" to match every year's evolving specifications, and never sold a penny's worth of any of it.

    Our hardware used ethernet for this, but I'm sure others used FDDI,
    because in those days, ethernet was still 10Mbps.

    --
    Lars Poulsen - an old geek in Santa Barbara, California

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lawrence D?Oliveiro@3:633/10 to All on Sun Aug 9 23:20:52 2026
    On Sun, 9 Aug 2026 15:39:06 -0700, Lars Poulsen wrote:

    Our hardware used ethernet for this, but I'm sure others used FDDI,
    because in those days, ethernet was still 10Mbps.

    Meanwhile, IBM was still pushing 4Mb/s Token Ring, which I think they
    claimed was just as good as 10Mb/s Ethernet. Because collisions.

    A colleague of mine told me about a demo Sun did, involving two pairs
    of its workstations sharing the same cable segment, with heavy data
    transfers going on between the two members of each pair. In other
    words, a scenario calculated to maximize the collision losses.

    Apparently they measured something like 98% bandwidth utilization. In
    other words, collisions and other overhead lost only 2%.

    Sun were also known for creating Ethernet interfaces that were right
    on the limits of the spec. Put them on the same cable as some cheapo
    (for the time) PC interfaces (not helped by MS-DOS driver
    limitations), and with the Suns running flat out, the poor PCs would
    be struggling to get a frame in.

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)