Apollo Computer

[Lawrence D'Oliveiro]

I was looking at the Apollo Archive <https://jim.rees.org/apollo-archive/>
to find out more about this former player in the turbulent Unix
workstation market. They offered some powerful graphics hardware, with
their own proprietary 3D API, almost I think on a par with SGI in its
early days. Another distinguishing characteristic was the highly
integrated networking among different workstation nodes.

They were founded in 1980, and initially their hardware ran a proprietary
OS called “Aegis” (which they described as “object-oriented”, with its own
proprietary GUI), on their own CPU architecture (described as “bit-
sliced”). (I think these were the products with 3-digit model numbers,
e.g. 460, 660.) Then within a few years they moved to Motorola 68020/68030
processors (and 4-digit model numbers, e.g. 3500, 4500). Also they started
using the name “Domain” for their software stack; I think initially this
just referred to the network system; then this became “Domain/OS” for the
whole OS, which offered both BSD and AT&T Unix as “personalities” (my
term, not theirs) on top of the core, in addition to the older Aegis
support. Later, I think, when Aegis was abandoned, the OS became “Domain/
IX”, which was a totally Unix-based system.

In 1988 or so, they introduced their first RISC-based machine, the Series
10000 “Personal Supercomputer”. This ran an architecture they called
“PRISM”, and was fabulously expensive -- up to a 6-figure price tag. Their
less-expensive “personal workstations” (with only a 5-figure price tag)
were still using Motorola processors, while most of the other Unix
workstation vendors were already moving large parts of their product
ranges to RISC.

Looking at brochures from this date, I still cannot see any mention of X11
support -- looks like they were still sticking to their own proprietary
GUI, when most of their competitors were quick to adopt the new open
standard for this.

Then, at some point, they got acquired by HP, no doubt as a result of
their struggles to stay competitive in a fast-changing marketplace. In the
above archives, in some update files grouped by year from 1991 onwards, I
see subdirectories labeled “a88k” and “m68k”. So it looks like, not only
were they still supporting (and therefore still selling?) machines based
on Motorola 68000-family processors, but they were now using Motorola’s
88000 RISC processor as well -- their fourth CPU architecture, by my
count.

Wasn’t PRISM good enough? Was it too expensive, maybe? From what I’ve
heard, the 88000 family weren’t particularly wonderful performance-wise,
which is why hardly anybody made use of them. So, after dragging their
feet over Unix and then X11 support, yet another in a series of
questionable strategic decisions from the company? Which is why, after a
few more years, it ceased to exist altogether.

[John Dallman]

Post by Lawrence D'Oliveiro
They were founded in 1980, and initially their hardware ran a
proprietary OS called _Aegis_ (which they described as
_object-oriented_, with its own proprietary GUI), on their own CPU
architecture (described as _bit-sliced_).

Wikipedia reckons they started with 68000, and the bit-slice CPUs were
their own implementation of 68000. They seem to have been keen to spend
money on hardware, since their method of handling page faults on their
68000-based machine was to have another 68000 to handle them. This wasn't
needed with their 68010 and later machines

<https://en.wikipedia.org/wiki/Apollo/Domain>

Post by Lawrence D'Oliveiro
Looking at brochures from this date, I still cannot see any mention
of X11 support -- looks like they were still sticking to their own
proprietary GUI, when most of their competitors were quick to adopt
the new open standard for this.

The comp.sys.apollo FAQ reckons they had X11 at some point:

<https://web.mit.edu/kolya/www/csa-faq.html>

Post by Lawrence D'Oliveiro
Wasn't PRISM good enough? Was it too expensive, maybe?

Too expensive, it seems. A "personal workstation" at up to $100,000 was a
very high price in the late 1980s. They sold about a thousand PRISM
machines:

<https://en.wikipedia.org/wiki/Apollo_PRISM#History>

John

[John Levine]

Post by John Dallman

Post by Lawrence D'Oliveiro
They were founded in 1980, and initially their hardware ran a
proprietary OS called _Aegis_ (which they described as
_object-oriented_, with its own proprietary GUI), on their own CPU
architecture (described as _bit-sliced_).

Wikipedia reckons they started with 68000, and the bit-slice CPUs were
their own implementation of 68000. They seem to have been keen to spend
money on hardware, since their method of handling page faults on their
68000-based machine was to have another 68000 to handle them. This wasn't
needed with their 68010 and later machines

That matches my recollection. There was a bug in the 68K that made it impossible
to recover from a page fault so Apollo came up with a kludge that froze the main
CPU and switched to the other one. Later 68K chips didn't have the problem so
one CPU was enough.

It is also my recollection that they waited a long time before switching from
their own system to Unix. Their system was pretty good but it didn't have all
that software or prople who knew how to write programs for it.

[John Dallman]

Post by John Levine
It is also my recollection that they waited a long time before
switching from their own system to Unix. Their system was pretty
good but it didn't have all that software or prople who knew how
to write programs for it.

My employers supported it, but gave up long before I joined them. There
are still people there who use "ale", the Apollo-Like Editor, and some
platform-specific code for Apollo in the product I work on.

John

[George Musk]

Post by John Dallman
are still people there who use "ale", the Apollo-Like Editor, and some

Is it something different than
https://texteditors.org/cgi-bin/wiki.pl?ARPUS_Ce
? Can't find it.

[John Dallman]

Post by George Musk

Post by John Dallman
are still people there who use "ale", the Apollo-Like Editor

Is it something different than
https://texteditors.org/cgi-bin/wiki.pl?ARPUS_Ce

It's different: one they wrote themselves in the late 1980s.

John

[Scott Lurndal]

Post by John Dallman

Post by John Levine
It is also my recollection that they waited a long time before
switching from their own system to Unix. Their system was pretty
good but it didn't have all that software or prople who knew how
to write programs for it.

My employers supported it, but gave up long before I joined them. There
are still people there who use "ale", the Apollo-Like Editor, and some
platform-specific code for Apollo in the product I work on.

We had a bunch of Apollo workstations at Burroughs in the 80's; they
were exclusively used for hardware description language development
for ECL gate arrays. They were eventually replaced with Solbourne
sun-compatable workstations.

[Malcolm Purvis]

Post by Lawrence D'Oliveiro
Looking at brochures from this date, I still cannot see any
mention of X11 support -- looks like they were still sticking to
their own proprietary GUI, when most of their competitors were
quick to adopt the new open standard for this.

Full X11 support arrived with their SR10 OS release around 1988,
which introduced their Domain/OS name and OS personalities.

I was a 4 year student at the University of NSW in Sydney
Australia when the Computer Science department replaced the
student VAXen and PDP11s with a large number of Apollo machines.
The remnants of their original OS caused great confusion to a
school who were more used to UNIX v7.

They were all decommissioned 4 years later. I bought one of the
more powerful workstations, full docs and OS tapes for AU$500
because nobody else wanted it.

Malcolm

[Johnny Billquist]

Gah! Horrible memories.

Their window system was the weirdest thing with the terminal window
having a special sub-window where you wrote your input. The point
pointer and the window cursor was actually the same, and things behaved
in very weird ways.

Aegis had some cool ideas though. The file system was networked, and you
had the machines a level above the root directory, and soft links could
contain environment variables that got expanded when you actually used
them, so you could have file system entries that pointed in different
ways depending on which machine or environment you were on.

But then Apollo was also very fond of using token networks, which were
*really* bad.

JOhnny

Post by Lawrence D'Oliveiro
I was looking at the Apollo Archive <https://jim.rees.org/apollo-archive/>
to find out more about this former player in the turbulent Unix
workstation market. They offered some powerful graphics hardware, with
their own proprietary 3D API, almost I think on a par with SGI in its
early days. Another distinguishing characteristic was the highly
integrated networking among different workstation nodes.
They were founded in 1980, and initially their hardware ran a proprietary
OS called “Aegis” (which they described as “object-oriented”, with its own
proprietary GUI), on their own CPU architecture (described as “bit-
sliced”). (I think these were the products with 3-digit model numbers,
e.g. 460, 660.) Then within a few years they moved to Motorola 68020/68030
processors (and 4-digit model numbers, e.g. 3500, 4500). Also they started
using the name “Domain” for their software stack; I think initially this
just referred to the network system; then this became “Domain/OS” for the
whole OS, which offered both BSD and AT&T Unix as “personalities” (my
term, not theirs) on top of the core, in addition to the older Aegis
support. Later, I think, when Aegis was abandoned, the OS became “Domain/
IX”, which was a totally Unix-based system.
In 1988 or so, they introduced their first RISC-based machine, the Series
10000 “Personal Supercomputer”. This ran an architecture they called
“PRISM”, and was fabulously expensive -- up to a 6-figure price tag. Their
less-expensive “personal workstations” (with only a 5-figure price tag)
were still using Motorola processors, while most of the other Unix
workstation vendors were already moving large parts of their product
ranges to RISC.
Looking at brochures from this date, I still cannot see any mention of X11
support -- looks like they were still sticking to their own proprietary
GUI, when most of their competitors were quick to adopt the new open
standard for this.
Then, at some point, they got acquired by HP, no doubt as a result of
their struggles to stay competitive in a fast-changing marketplace. In the
above archives, in some update files grouped by year from 1991 onwards, I
see subdirectories labeled “a88k” and “m68k”. So it looks like, not only
were they still supporting (and therefore still selling?) machines based
on Motorola 68000-family processors, but they were now using Motorola’s
88000 RISC processor as well -- their fourth CPU architecture, by my
count.
Wasn’t PRISM good enough? Was it too expensive, maybe? From what I’ve
heard, the 88000 family weren’t particularly wonderful performance-wise,
which is why hardly anybody made use of them. So, after dragging their
feet over Unix and then X11 support, yet another in a series of
questionable strategic decisions from the company? Which is why, after a
few more years, it ceased to exist altogether.

[Lynn Wheeler]

Post by Lawrence D'Oliveiro
Wasn’t PRISM good enough? Was it too expensive, maybe? From what I’ve
heard, the 88000 family weren’t particularly wonderful performance-wise,
which is why hardly anybody made use of them. So, after dragging their
feet over Unix and then X11 support, yet another in a series of
questionable strategic decisions from the company? Which is why, after a
few more years, it ceased to exist altogether.

POWER/RIOS (six chip) chipset didn't support multiprocessor cache
coherency so scale-up was cluster. Executive we reported to (when doing
HA/CMP),
https://en.wikipedia.org/wiki/IBM_High_Availability_Cluster_Multiprocessing

then went over to head up Somerset (i.e. AIM; apple, ibm, motorola)
single-chip processor and I somewhat characterize as adding motorola 88k
risc multiprocessor cache coherency ... then can have large scalable
clusters of multiprocessor systems (rather than just clusters of single
processor systems)

https://en.wikipedia.org/wiki/AIM_alliance
https://wiki.preterhuman.net/The_Somerset_Design_Center
https://en.wikipedia.org/wiki/IBM_Power_microprocessors#PowerPC
https://en.wikipedia.org/wiki/Motorola_88000
In the early 1990s Motorola joined the AIM effort to create a new RISC
architecture based on the IBM POWER architecture. They worked a few
features of the 88000 (such as a compatible bus interface[10]) into
the new PowerPC architecture to offer their customer base some sort of
upgrade path. At that point the 88000 was dumped as soon as possible