Jumping in late because the list blew up so badly on this topic. Yes, others
have already commented on these things, but I'll add my US$0.02 worth anyway.
> Date: Thu, 9 Mar 2023 16:00:05 -0600
> From: Steve Lewis via cctalk <cctalk@classiccmp.org>
> Actually, to answer my own question: if "main frame" refers to the actual
> framing... well the PDP-1, PDP-10, PDP-10 were minicomputers and still
> required a lot of metal "framing" to set up. So, can't they be considered
> mainframes?
Don't be fooled by the naming convention of a single computer manufacturer who
was getting around GAO rules about computers. Before the coining of the term
"minicomputer", most systems from Digital Equipment Corporation were classified
as "small computers". This nomenclature covers the PDP-1, the PDP-4/7/9/15,
and the PDP-5/8 (-8, -8i/l, -8e/f/m, -8A).
However, the PDP-6 was marketed into the same customer space as IBM and the
Seven Dwarfs (Burroughs, Univac, CDC, NCR, Honeywell, RCA, and GE), which
became the BUNCH after RCA and GE sold their computer divisions to other
members of the group. The PDP-6 were advertised as a multiuser system which
featured built-in timesharing (an extra cost add-on feature of some of the
others at the time) as well as batch processing. It supported large disks and
drums, high speed printers and card reader/punches, and strings of magnetic
tape drives. It required a staff of professional operators and systems
programmers to run.
By all measures, the PDP-6 was a 1964 mainframe computer.
The PDP-10 was a reimplementation of the PDP-6 using a safer technology (the -6
turned out to be fragile, and nearly bankrupted the company). The engineering
for the -10 was done as a skunkworks project, and the mainframe nature of the
system was hidden from Ken Olsen, the CEO/founder, but customers were not
fooled and recognized it as a mainframe.
> (another notion is that mainframes are "multi-user" -- most early
> microcomputers were not multi-user, as they just barely supported the needs
> of one user; I'm not sure if the very first minicomputers were multi-user?)
Early mainframes were single user, in the sense that only one program could be
run at a time. Spooling of jobs was invented to alleviate the time "wasted" in
having operators set up and reconfigure jobs, but it was still a one-at-a-time
thing. Timesharing was an extra cost add-on based on research systems at
places like MIT.
The very first minicomputers were indeed single-user--but see below.
> The term minicomputer has always been awkward to me -- "mini" in my head
> just means something smaller than me, which most minicomputers aren't (but
> they are much smaller than a building). But to say "mainframe" when
> showing a minicomputer then necessitates some explanation... Can't win :(
The term "minicomputer" was marketing speak: The first computer to receive the
appellation was the PDP-8/e, which was the third generation of the PDP-8 family
(where the PDP-5 is "generation zero"). The first generation PDP-8 fit into
the back of a VW convertible (a famous marketing photo); the third would fit on
the passenger side front seat. It came out at a time when the miniskirt was in
full bloom, and everything in the marketing world was "Mini! Mini! Mini!"--even
when it wasn't.
BTW, the PDP-8/i (second generation) *did* have an extra cost option to be a
multiuser timesharing system, with an operating system called TSS-8. It was
created by the engineers who built the PDP-10, because they wanted small system
users to have access to the cool features of that mainframe. (I was told this
by one of the designers of the PDP-10, Bob Clements, who also worked on TSS-8).
Rich
