9 Aug
2026
9 Aug
'26
1:51 p.m.
Hello everyone, (cross posting to TUHS)
I have an AT&T WE 321SB VMEbus single-board computer en-route to me.
It's sort-of a 3B2/3B5 on a VME card. It has WE32100 CPU, WE32106 MAU,
and a WE32101 MMU all running at 18 MHz. This seems to put it roughly
in-between the V2 and V3 3B2 architecture era. I believe it was
initially offered by AT&T as a WE32K developer system, but I'm not super
familiar with it's history. Once it arrives, I'll of course upload the
ROM dumps to several archives. The board does not come with an intact
filesystem or installation media. I am reaching out to see if anyone in
the community has or knows the whereabouts of any media or documentation
for the board or UNIX System V/VME software for the WE32K. I've seen
references to "WE 321EB Microprocessor Evaluation Board User Manual"
(AT&T Select Code / Catalog Number: CAT 451-019). But it also seems to
be missing from history. There is also photos of the install media
containing:
WE 321SB-VSU Software Utility Programs
VME Device Driver Utilities / VME Dual VSMD-32 SMD Driver
VME Device Driver Utilities / VME Ironics IV-1624 IS-I/O Driver-Source
UNIX System V/VME / Essential Utilities-Disk 2
UNIX System V/VME / Essential Utilities-Disk 2
UNIX System V/VME / Startup Disk
If you can help in any way in preserving another rare WE32K artifact,
please let me know.
Thanks,
Alan Hightower
10 Aug
10 Aug
3:56 p.m.
I am being slammed by email today so I will have to be brief.
I was hired in 1987 by Mizar, Incorporated, to bring up AT&T
SVR4 v2 Unix (sic) for them to sell as a VMEBus product. IIRC my
development machine at that time had a 321SB board in it. My
understanding is that the only difference between it and the 3B2
machines was the bus interface. VMEBus was going to be "the future"
of expansion buses for industrial computers at that time.
Even by the standards of the time this CPU was underpowered. IIRC
we made our own boards based on it. (VMEBus master boards made by
manufacturers who did not specialize in VMEBus, as we did, tended not
to be well-designed, and only implemented a minimum of features.
Their multiple bus master support was non-existant, for instance.)
The product did not sell. After a year they switched me over
to the design of the next "future" processor, a sparc32 board.
That did not sell well, either.
Then we tried to sell the idea of MIPS-based boards to interested
customers but not much happened.
During that time Mizar was making most of its money off the 68k
product lines. The 68000-, 68010-, and 68020- VMEBus machines
were still selling well. We were selling these machines either
to run bare-metal, PDOS, or vxWorks. (The latter was what I was
assigned to the whole time.)
Older people will recall that Motorola "lost its mojo" around that
time and were having trouble scaling up the 68040 to faster clock
speeds (this is because much of their technical design team walked
out the door to start Ross Technology.) So we were trying to figure
out what we could offer to stay competitive.
tl;dr: we didn't. I left at the end of 1989 after running my mouth
to management too much.
I can go on a treasure-hunt to see what I still have from those days
but it won't be this week. I do have manuals and maybe even a board
or two.
AFAIK no one else ever even got AT&T-based processors out the door
at all.
mcl
11 Aug
11 Aug
10:24 a.m.
On Mon, Aug 10, 2026 at 3:56 PM Mark Linimon via cctalk
<cctalk(a)classiccmp.org> wrote:
I am being slammed by email today so I will have to be brief.
I was hired in 1987 by Mizar, Incorporated, to bring up AT&T
SVR4 v2 Unix (sic) for them to sell as a VMEBus product. IIRC my
development machine at that time had a 321SB board in it. My
understanding is that the only difference between it and the 3B2
machines was the bus interface. VMEBus was going to be "the future"
of expansion buses for industrial computers at that time.
Even by the standards of the time this CPU was underpowered. IIRC
we made our own boards based on it. (VMEBus master boards made by
manufacturers who did not specialize in VMEBus, as we did, tended not
to be well-designed, and only implemented a minimum of features.
Their multiple bus master support was non-existant, for instance.)
The product did not sell. After a year they switched me over
to the design of the next "future" processor, a sparc32 board.
That did not sell well, either.
Then we tried to sell the idea of MIPS-based boards to interested
customers but not much happened.
During that time Mizar was making most of its money off the 68k
product lines. The 68000-, 68010-, and 68020- VMEBus machines
were still selling well. We were selling these machines either
to run bare-metal, PDOS, or vxWorks. (The latter was what I was
assigned to the whole time.)
I think Motorola did the lion's share of VME through longevity,
although some early UNIX workstations like Sun2/3 and SGI also used it
as an expansion bus. I grew up down the street from Motorola Computer
Group and have a small collection of oddities. The MVME boards are
interesting UNIX machines (System V/68K and 88K) and still supported
by NetBSD (68k).
Veering slightly off topic, VME was somewhat displaced by CompactPCI
(and to a lesser extent ATCA/uATCA), although it probably will outlive
everything else (it shows up in weird places with long lifetimes like
ASML and other semiconductor equipment, particle accelerators, M1
Abrams tanks). These days it seems like the nearest neighbor is a
SOM. Maybe the cost/difficulty of creating custom boards has just
dropped low enough.
Older people will recall that Motorola "lost its mojo" around that
time and were having trouble scaling up the 68040 to faster clock
speeds (this is because much of their technical design team walked
out the door to start Ross Technology.) So we were trying to figure
out what we could offer to stay competitive.
tl;dr: we didn't. I left at the end of 1989 after running my mouth
to management too much.
I can go on a treasure-hunt to see what I still have from those days
but it won't be this week. I do have manuals and maybe even a board
or two.
The manuals for VME boards tend to be gold, usually full details on
the bring up and register space, I guess because the customers were
sophisticated.
AFAIK no one else ever even got AT&T-based
processors out the door
at all.
I haven't heard of other uses of the WE32k. There were certainly many
choices in that era.
mcl
13 Aug
13 Aug
9:09 p.m.
Mizar sold a WE32K product? Never saw that one. Wild.
On Tue, Aug 11, 2026 at 1:25 PM Kevin Bowling via cctalk <
cctalk(a)classiccmp.org> wrote:
I haven't heard of other uses of the WE32k. There were certainly many
choices in that era.
Not the WE32K processor, but oddly enough the WE32104 DMA controller shows
up in a number of MC68K designs (and perhaps others). Heurikon as an
example. I assume that's because Motorolas discrete DMA controllers
covered only 24-bits of address.
KJ
14 Aug
14 Aug
1:29 a.m.
That's not a 3B2, My cube neighbor at metheus did a DRV-ll style
parallel card. It was long and painful.
Last place I know where he worked was Rogers instruments as the senior scientist
Maybe he has some old stuff in a box. I sold all my interesting stuff
to Josh. Well, mostly
Jim.
On Thu, Aug 13, 2026 at 9:09 PM Ken Seefried via cctalk
<cctalk(a)classiccmp.org> wrote:
Mizar sold a WE32K product? Never saw that one. Wild.
On Tue, Aug 11, 2026 at 1:25 PM Kevin Bowling via cctalk <
cctalk(a)classiccmp.org> wrote:
I haven't heard of other uses of the WE32k. There were certainly many
choices in that era.
Not the WE32K processor, but oddly enough the WE32104 DMA controller shows
up in a number of MC68K designs (and perhaps others). Heurikon as an
example. I assume that's because Motorolas discrete DMA controllers
covered only 24-bits of address.
KJ
27
days inactive
32
days old
5 comments
5 participants
participants (5)
-
Alan Hightower -
Jim Davis -
Ken Seefried -
Kevin Bowling -
Mark Linimon