Interesting photos of a computer graphics lab from 1968

[Marc Howard]

Those photos take me back.  I worked at the graphics research lab at NC
State and we were still using an Adage AGT-30 in the 70's.  Here's some
info about the lab with more pictures:
https://www.graphics-history.org/ncsu/index.htm

Hard to tell from the picture but that is probably a model 30 as well.  The
AGT-10 was 2D only, the 30 added 3D and the 50 added a hardware character
generator.

Very interesting architecture.  Logic levels were +/- 2.7 volts, used
discrete, non-saturating logic.  30 bit word (6 bit characters) size, 32K
words maximum.  1's complement.  Basic cycle time was 4 usec  The ALU was
made from DTL logic and had hardware multiply (15 x 15 in 6 usec).

Very RISC style architecture, virtually all instructions were "transfers"
from one of 16 sources and one of 8 destinations.  Although you could code
a memory to memory instruction it wouldn't actually do anything.

Also in each transfer were two transfer type bit which allowed the dual
rail transfer bus (similar to JK flip flop inputs) to do a true,
complemented (which is negation in 1's comp), OR'd or AND'd transfer to the
source.  And on top of that were Index and Indirect modifier bits for
memory access.  Two more bits allowed for 0, 1, 6 or 15 bit rotation of the
destination operand as well.

The Index and Indirect bits were interesting in that they occured
autonomously in the memory subsystem.  Indirection would go on until an
access was made without the indirect bit being set.  If you coded an
instruction like "JUMP'I ." indirection was infinite and couldn't be
stopped even if there was an interrupt.  Literally power-off was the only
escape.

The disk drives were made by ICL (Twin Disc) and were the only removable
disk packs I've ever seen that spun in the vertical axis. Tape drives were
usually 7 track although 9 track was available as well.

Interrupts were accomplished by jamming an instruction on the data bus.
For example the 7 track drive used two different instructions.  For the
first 4 characters read off tape the buffer register was commanded to
rotate 6 bits and OR in data on the data bus.  On the interrupt five an
instruction OR'd in the last byte and indirectly indexed a fixed pointer
location thus putting 30 bit data in a buffer in memory.  Zero overhead
transfers, much like the 3 cycle data break in the PDP-8.

That's just the digital part.  On the 30 or 50 there was a 19" rack full of
15 bit signed discrete multiplying DACs.  The DACs were organized in a 4 x
3 array which allowed for vectors to be drawn with 3D depth queuing (more
distant parts of a display had perspective as well as indesity reduction).
The beauty of the interrupt subsystem was again that data could be fetched
as needed by the display hardware with no software involvement until an
end-of-list was detected or a light pen interrupt occurred.

Achilees heel was displaying text.  You generally set the editor to only
show about 6 lines of context due to the refresh flicker.  BTW, the editor
was modeless, like TECO.  Unfortunately, it edited 6 bit characters only,
i.e., no lower case.

The software was very advanced.  Fortran IV compiler with inline assembler
support and complex numbers.  Greate editor. You scrolled through a file
using foot pedals.  That was a great feature.

 Adage was located in Boston and MIT/DEC influence shows up in their
hardware and software.

The assembler is the only one I've ever seen that allows you to modify the
assembler executable* DURING ASSEMBLY*.  It did this by starting out with
all of the assembler variable addresses pre-loaded into the symbol table at
the start.  Normally, you would use the word "EXPUNGE" to zero out the
symbol table before assembling.  But if you didn't you could define code
that overwrites the loaded assembler code.  I was able to use this feature
to retarget the assembler to generate code for a TMS9900 processor board I
designed and built for myself.  Think of it.  An assembler for a 30 bit 1's
complement processor was modified at run time to generate code for a 16
bit, two's complement machine.

Also in that lab I inherited a non-working Ambilog-200.  The predecessor to
the AGT series but 2D only.  Here's a picture of the console:
https://www.graphics-history.org/ncsu/ambilog200_small.jpg
Selectric IO typewriter and all the registers in your face!  That was my
first "PC".

I miss that machine.  I'd love to sim it but the code and schematics are
gone forever.  Actually I think CHM has a box of stuff but I can't find a
link to it anymore.

BTW, The AGT made a movie appearance in John Carpenter's pre-Star Wars sci
fi cult classic "Dark Star".  Youtube has the whole movie (
https://www.youtube.com/watch?v=ocse-0bBfo8 ).  If you go to about the
52:30 mark you'll see a brief appearance of the AGT control panel doing
what looks like loading in the first stage bootloader from paper tape.  The
110 baud paper tape took an interrupt on each bit, ORing in the next bit
and on the 30th bit loading the value into the next memory location.

Marc








On Thu, Jun 24, 2021 at 10:45 AM Mike Begley via cctalk <
cctalk at classiccmp.org> wrote:

> A friend of mine collects old photos, and sent me a link to this set he
> just recovered from film.  He doesn't know much about the provenance, but
> it might be interesting to figure out what we can about them.
>
> The link is here:
>
> https://www.espressobuzz.net/Found/GeorgeClark/GraphicsLab/
>
> The film roll was dated June 1968, which was a month before I was born.
>
> The most interesting (to me) is the photo of the Adage Graphics Terminal.
> I'd never heard of the company before, but it looks like they were one of
> the many tech companies that sprang up in the MIT & Harvard orbit in the
> 60s and 70s.  There's a small amount of info about the company on
> Wikipedia:
> https://en.wikipedia.org/wiki/Adage,_Inc.
>
> Also in the same photo, to the right of the terminal is a silver box that
> might be an early joystick.  It's hard to say.
>
> The pictures of the graphics terminal output is pretty cool, given that
> it's probably really pushing state of the art at the time.  Also, they're
> in color, which was still not terribly common at the time.  The only few
> photos I was able to find of Adage terminals are all black and white.
>
> Other recognizable hardware are a couple of ASR33 teletypes (one of which
> was rebadged as Adage), and some tape drives, the manufacture I don't
> recognize.  Everything else, I pretty much can't make out what any of it
> is, but perhaps someone recognizes the particular layout of the
> blinkinlights?
>
> He was told that the pictures were taken "somewhere in the northeast".  I
> suspect Boston, and therefore probably MIT or maybe Harvard.  On the same
> roll were some pictures of "a colonial ship with lots of cannons", which I
> suspect was the USS Constitution in Boston Harbor, but I haven't seen those
> pics yet.  I wonder if the tile pattern on the floor is distinct enough or
> recognizable?
>
> Anyway, it's an interesting set of archival photos, and I figured someone
> here might find them interesting or might recognize more than I do.
>
> -mike
>
>