J. Peterson wrote:
The 4081: Full "workstation" with a 16 bit
mini built into a
side cabinet, with two cartridge disk packs (roughly RK-05
sized, but the ones that dropped down into a drawer. The CPU
was roughly equivalent to an PDP-11/05, but was some
off-brand make (DG? Perkin-Elmer?) It had a 4014-style tube
driven directly by the CPU. It ran a proprietary RT11-like OS
called "GOS". Reportedly the pet project of the son of one
of the company founders.
The Tektronix 4081 used an OEM Interdata 7/16 CPU board set. The disk
drives were, IIRC, made by Wangco (a different company than Wang
Laboratories). The drives had one fixed platter, and one removable
platter, like an RK05F, but definitely different. Can't remember the
exact capacity, but 10MB per-platter seems to ring a bell. Most of the
systems I worked with had two of these drives. They used a disk
cartridge that was similar to that used in a DEC RL01/RL02.
The display wasn't driven "directly" by the CPU. There was a secondary
graphics processor (designed by Tektronix, using AMD bit-slice
technology, IIRC) that interpreted a display list placed into a movable
segment of main memory. The CRT had write-through capability, and due
to the power of the display list processor, you could do pretty darned
good dynamic vector graphics with the thing.
The backplane of the machine was Tek-designed, but general bus structure
followed the Interdata model. Our group had a hardware engineer in it,
who designed a RAMDisk that sat on the bus and used DMA to transfer back
and forth between main memory. The bus was rather weird, and they had
a heck of a time (even with help from the 4081 group) getting the board
to talk. Once it worked, some hacks were made to the OS so that
swapping (it was not a VM operating system) would be done to the RAMDisk
which really sped things up. The OS and applications were also modified
to store temporary files out on the RAMDisk.
The OS was indeed called GOS, and was very similar to RT-11. From a
programmers standpoint, the Interdata CPU was much more "IBM"-360-like
than "DEC PDP-11"-like. I worked for a group within Tektronix called
"CAD Development", which was an internal group that designed and wrote
CAD software for Tektronix engineering use. We had a couple of 4081's
in the group. They had written software for the 4081 that did a
respectable job of schematic capture and circuit board design (no
auto-route though...that job was left to the Control Data Cyber 73 and
an in-house developed routing package called "PIRATE" (no connotation --
the routing software was original), written mostly in Fortran 77 on the
Cyber. The 4081 could also generate artwork tapes for a Gerber
photoplotter, as well as NC tapes for circuit board drilling/routing.
There was a Pascal compiler for the 4081 that I ended up writing some
code in. I never cared much for Pascal, but it worked, and it was a lot
more productive than writing in assembly (which I also did some of).
The disk drives were mounted in the base of the cabinet, and were prone
to collecting all kinds of dust and fuzz, which, if not serviced
regularly, would inevitably lead to disk crashes. Fortunately, the
removable disk seemed more prone the crashes than the fixed disk (makes
sense), but when the fixed disk crashed, it was pretty ugly. The
machine had a ROM bootloader that allowed booting from a 4051-like
cartrige magtape drive, or disk. Installing the OS took a LONG time
from cartridge tape. I believe that machines were also
able to directly
interface to some kind of 9-Track magtape drive (reel-to-reel), and
may
have been bootable from said device.
I know of no currently existing 4081's, but who knows what may be out
there. It was definitely an interesting machine.
I also heard the rumor that the machine was indeed a pet project of a
son of some very-high-ranking Tektronix executive. Can't substantiate
it, though.
The Tektronix 4081 was used to generate the vector graphics for film of
the full-screen graphics used in the original "Battlestar Galactica" TV
show. The 4051's with the write-through modification were not fast
enough (even when coded in machine code) to generate complex enough
graphics, which is why the 4081 was used, because its display list
processor made much more complex refreshed raster graphics possible.
The 4081 was not actually used on-set, the on-set computers were Tek
4051's. Along with the 4051's there were wall-fulls TM-500 test
equipment.
Rick Bensene
Tektronix Employee from 6/1977 through 9/1990
The Old Calculator Web Museum
http://oldcalculatormuseum.com