.png)
4 months ago
10
Summer 2025
Amiga lore is full of exciting tales. Many of them are retold to demonstrate how the incompetence of Commodore's management destroyed a platform that, by rights, was destined for success. Coulda, shoulda, and the Amiga woulda risen as rightful ruler of all other computer platforms, forever and ever. Amen.
One of those stories is about how Sun Microsystems allegedly showed interest in the Amiga 3000 during the early 1990s. It's a classic Amiga anecdote, usually recounted without much reflection, and one I've certainly helped perpetuate.
Alas, the more I think about it, the less it adds up. Fact or factoid? Let's speculate!
Historical Sources
The Amiga 3000 was launched in 1990. Featuring on-board SCSI, several high speed expansion slots, a 25 MHz Motorola 68030 CPU and a 68882 FPU clocked at the same speed, it was certainly a very competent Amiga model. During 1991, Commodore launched a rebadged version dubbed Amiga 3000UX, which shipped with Commodore's own port of UNIX System V, Release 4 (or SVR4 for short).
At the time of writing this text, the Wikipedia article on Amiga UNIX refers to a possible Sun deal as "unsubstantiated rumors (...) presented in various online venues." - but that's Wikipedia. Apart from rumors presented as such in the 1990s computer press, there are two other sources regarding a deal between Sun and Commodore.
The first is Dave Haynie, a hardware engineer at Commodore who - among other things - worked on the Amiga 3000. In 1994, Haynie filmed, edited and released the video documentary The Deathbed Vigil. Interspersed between footage of Commodore's offices and staff (including people who worked on the Amiga UNIX port) are various informational screens of text. One of those states the following: "Sun Microsystems had planned to OEM the A3000UX as their answer to low-end, 680x0 based UNIX machines, a complement to their new high performance SPARCs. Unfortunately, the upper management at Commodore killed the deal."
The second source is another Commodore engineer, AmigaOS developer Bryce Nesbitt. In Brian Bagnall's 2019 book "Commodore: The Final Years", Nesbitt is interviewed and recounts how Commodore's UNIX department approached Sun in 1991 and suggested they'd license Amiga UNIX for their low-end 68030 machines. Sun allegedly responded positively, but the negotiations fell through when Commodore manager Medhi Ali started meddling in the deal and demanded more money. The curious thing is that it could be perceived as if Haynie is corroborating this version in Bagnall's book, but it's hard to tell exactly and seems to be based on an out of context quote used in conjunction with Nesbitt's story.
Discrepancies
Interestingly, the two stories differ in some key details. Haynie - a hardware guy - talks about licensing the hardware. Nesbitt - a software guy - talks about licensing the software. In Haynie's version, Sun approaches Commodore; in Nesbitt's, it's the other way around. As far as I can surmise, neither Nesbitt nor Haynie were directly involved with Commodore's UNIX department. In fact, several Commodore engineers seemed to think it was a pointless waste of energy: "A total distraction that didn't go anywhere" is Nesbitt's own summary of the UNIX effort, according to Bagnall's book. (Full disclosure: I agree with Nesbitt in this assessment, which may of course affect my judgment.)
I have neither intention nor desire to call former Commodore engineers liars, and I don't believe they are. Nevertheless, human memory is a fickle thing and in a game of corporate Chinese whispers, it's not unlikely for tidbits of information, however initially accurate, to grow into something else in the minds and hearts of passionate but disgruntled employees.
Personally, I think the whole thing is riddled with circumstances that make very little sense.
The Road to Amiga 3000 UNIX
Commodore had been circling UNIX for a long time before the release of the A3000UX. In 1985, they announced the Commodore 900, a Zilog Z8001-based machine running Coherent, a then popular UNIX-like operating system. The project was halted at the prototype stage, though working machines had been demonstrated at the 1985 Hanover Fair (which later became CeBIT).
After the C900 there seems to have been a UNIX workstation hiatus at Commodore, until their announcement of the A2500UX in 1988. This machine was based around an Amiga 2000 fitted with a 68020 accelerator card. It was demonstrated at both the 1988 and 1989 CeBIT fairs, but "once again the company failed to put a price or delivery date on the box" according to Unigram/X on March 25, 1989. The same article in Unigram/X also mentions plans for a 68030-based A2500UX (the 68030 is essentially a 68020 with an integrated MMU and larger caches). It's a bit unclear to me when or even if the A2500UX was ever officially released to market - in March of 1990, Swedish Datormagazin for example reports it as being yet unreleased. Still, specimens clearly exist among hardware collectors, which, granted, is true for various known Amiga prototype models as well.
The A3000UX, however, was undoubtedly officially released. In the February 4-8 1991 issue of Unigram/X, we can read the following regarding the product launch: "Commodore’s Unix manager Paul Calkin, who said Open Look [Sun's GUI environment, my remark] was picked because of its programming consistency and superior number of applications, suggested that there would be some sort of joint marketing done with Sun, but was not specific about how that would work out." According to Amiga World Magazine in April 1991, "Sun MicroSystems showed an A3000UX in its Open Look booth" at the Dallas Uniforum Show that same year - which is likely to have been what said joint marketing amounted to.
Hardware Hurdles
It seems highly improbable to me that Sun would have shown any specific interest in the A3000UX. Sun did indeed have a line of 68k machines, but their main focus at this point in time was the SPARC architecture. They had launched their Sun-4 SPARC machines already in 1987. In 1989, they launched the SPARCstation 1. At the same time, they also launched an updated line of Sun-3 machines (originally from 1985) sporting the 68030 CPU. According to MIPS Magazine, issue 6/1989, the introductory price for a diskless Sun-3/80 with a 20 MHz 68030, 4 megs of RAM and a monochrome screen was $5,995. Hence, by the time the A3000UX was launched, Sun already had a low-end 68k workstation - and an aging one, at that. History shows us that these machines were to be the last 68k machines ever made by Sun.
The base A3000UX sold for $4,998 according to the December 1991 issue of UnixWorld. For that price you got a 25 MHz machine with 1+4 megs of memory (the 1 meg essentially being for sound and graphics), a 100 meg SCSI hard drive and an Ethernet card - similar to the Sun-3/80, but without the latter's 17" monochrome screen. A fully equipped machine - with 1+8 megs of RAM, 200 meg hard drive, Ethernet, 1024x768 256-colour graphics and a 14" monitor - sold for $7,713.
The same UnixWorld review compares this to Sun's SPARCStation IPC, which it claims could be bought for $6,995 with 8 megs of RAM, a 200 meg hard drive, a 16" monitor and of course a much faster SPARC CPU. This, however, seems to be a misprint or perhaps misread on the reviewer's part. The Sparcstation IPC model referenced in the review was launched at $9,995 in 1990, and it seems unlikely to have dropped a full $3,000 - almost a third of its original price - in a mere year.
Similar "cheap" workstations such as the Apollo DN2500 came with specific caveats: the base model for under $4,000 boasted a high resolution 15" screen and a 20 MHz 68030, but - like the cheapest Sun-3/80 - no hard drive. According to BYTE issue 1, 1990, the DN2500 could be turned "into a stand-alone system by adding an internal 200-megabyte hard disk drive, with the operating system and utilities, for roughly $3,000 more." That's suddenly a $7,000 system - but with half the RAM and without the high-res 8-bit graphics offered by Commodore. It's also worth mentioning that Apollo's Domain/OS wasn't actually a UNIX, but merely offered a compatibility layer.
Hence, writing the A3000UX off as not being price/performance competitive isn't entirely fair. Sure, in the world of expensive UNIX workstations, it was middling. The base model came without a monitor and its 4 megs of RAM was starting to feel a bit cramped at the time. But unlike other workstations, it could run X in a monochrome 640x400 resolution by adding nothing but an off-the-shelf VGA screen. For an institution like Virginia Tech - Commodore's flagship A3000UX customer - this was likely one of the cheapest options available at that time if you wanted to run a modern, official UNIX on hardware that was reliably available for bulk purchase and came with a full factory warranty.
On the flip side, Motorola had released their 68040 CPU in 1990, providing a substantial performance increase compared to the 68030. Sun was candid about not planning to adopt it, instead focusing on SPARC. Their 68030 line was most likely intended as a way to placate legacy customers, while weaning them off the 68k architecture by offering trade-in deals for upgrading to newer machines. Did they really need another "complement" to this product line? If they'd had any real interest in reviving their 68k offering, it would presumably have been easy for them to, at the very least, put a 25 or 33 MHz 68030 into the Sun-3/80, and offer a disk-enabled model through various discount programs.
Workstation vendors that committed to the 68k range, such as HP and NeXT, had already launched 68040-based machines. Commodore, meanwhile, had yet to produce a 68040 accelerator card, which meant the upgrade ladder for their UNIX offering was uncertain. In the end, Sun won out: just a couple of years later all major workstation vendors had switched to RISC CPUs.
While offering modest UNIX performance, the Amiga 3000 was a beast compared to its older 7 MHz siblings, and a highly capable broadcast production workstation. The features it offered in this department, such as genlocking video and 8-bit stereo PCM sound, wasn't typically on demand among entry-level UNIX customers and probably offered little competitive edge in that department. People interested in video production, on the other hand, were used to appliance-like machines such as Quantel's Paintbox. They weren't particular about running UNIX - they wanted a Video Toaster and access to the many great graphics applications available for AmigaOS.
Given all of this, neither the suggested timeline nor the possible incentives add up. Why would Sun approach Commodore when they already had the Sun-3/80, a similar machine in roughly the same price bracket? Why launch a completely new 68030 model in 1991, when they'd already done so two years prior? And why suddenly re-commit to 68k when they explicitly wanted to focus on SPARC? It just doesn't make sense - not that business decisions always do.
Software Shenanigans
During the late 1980:s, Sun and AT&T worked together on a new version of UNIX - the fabled SVR4. In 1988, Sun announced that this was the future and that their next operating system would be SVR4-based. In reality, they kept on releasing new versions of SunOS 4.x, based on 4.3BSD. I'm not sure if this led to any specific complaints: SunOS was well established, Sun was a popular and successful vendor, and there was likely a substantial software library readily available. It seems just as probable that Sun didn't feel any particular need to rush things: low-effort backwards compatibility has always been a selling point, especially in the world of very expensive computers.
According to UnixWorld's Amiga 3000UX review in their December 1991 issue, Commodore's UNIX was the first port of SVR4 for Motorola's 68k CPUs. In theory, this could well have piqued Sun's interest. After all, their own machines didn't yet run the version of the OS they themselves had announced already in 1988.
But, once more, it seems the timeline just doesn't add up. In fact, I think this version of events is even more unlikely than any hardware licensing.
The porting effort wouldn't have been trivial, because apart from the CPU, the architectures were completely different: it wasn't simply a matter of compiling Commodore's sources on a Sun machine and call it a day. There was also the question of possible porting efforts needed for the existing SunOS software library. After all, what point is a 68k Sun computer if it can't run your 68k Sun programs?
The low-end 68k machines intended as a target for the port had been launched two years earlier. They were an outdated offering built around a CPU architecture Sun was phasing out - and they already ran UNIX. Sure, it wasn't SVR4, but nothing is as cheap as a popular, working product you already have. Why spend energy on a completely new OS for a cul-de-sac range of machines? And why promote these machines as a platform for their most modern OS when they reasonably wanted customers to invest in SPARC instead?
Solaris 2.0, Sun's fully SVR4-based UNIX, was announced in September 1991 and released in June 1992. If Commodore approached Sun as an effort to save Amiga UNIX in 1991, this would likely have occurred at a time when Sun knew they'd have their own SVR4 port ready within a year. Starting yet another porting effort at that point, especially to a CPU architecture that was being phased out, sounds like a waste of time and energy.
A summarized timeline of events
| 1985 |
Sun launches their Sun-3 line of 68020-based workstations. Commodore announces the Commodore 900, a UNIX workstation cancelled in the prototype stage. |
| 1987 | Sun launches their SPARC-based line of Sun-4 workstations. |
| 1988 |
Commodore announces the 68020-based A2500UX workstation. Sun and AT&T announces UNIX SVR4. |
| 1989 | Sun launches the SPARCstation 1 and 68030-based versions of their Sun-3 machines. |
| 1990 |
Commodore launches the 68030-based Amiga 3000, their first real
high end machine. Sun launches the cost-reduced SPARCstation IPC. |
| 1991 |
Commodore launches the A3000UX, the first 68k-based workstation
with a complete UNIX SVR4 distribution. Sun announces Solaris 2.0, a fully SVR4-based operating system for their SPARC machines. Sun allegedly expresses interest in licensing some or all of the A3000UX product offering. |
| 1992 |
Commodore releases Amiga UNIX 2.1, the last version
before discontinuing the product. Sun releases Solaris 2.0, their first SVR4 UNIX, exclusively for the SPARC architecture. |
The Truth?
In retrospect the A3000UX was cool, but ultimately doomed from the start. Acorn, Atari and Apple also tried but failed to muscle in on the UNIX workstation market in any meaningful way. Since Amiga fanboy honor is at stake, it's tempting to blame the failure of Amiga UNIX on Commodore's mismanagement.
We can see the two different Commodore engineer accounts either as proof that something had, indeed, been going on with Sun. We can also see the rather fundamental differences in the stories as proof that they're nothing but hearsay.
Could Haynie's version be true - that Sun wanted to license the A3000UX, despite having launched a low-cost 68030 line two years prior, and against their express desire to focus on SPARC? It seems highly unlikely to me, but that doesn't mean I'm right.
Could Nesbitt's version be true - that Sun wanted to license Amiga UNIX for their aging Sun-3 line, despite phasing out 68k, having a well-established platform in SunOS and an SVR4 UNIX for SPARC in the works? Again: improbable, but not unequivocally impossible.
Could both versions be sort of true - that Sun approached Commodore regarding one or both technologies, and Commodore botched the deal? I guess - but then the same caveats as above would apply.
Could Commodore have approached Sun, asking them to OEM the A3000UX or their SVR4, only to have Sun politely turn them down? If anything, this sounds more likely - and such an endeavour could very well have warped into a story about how Commodore's management screwed everything up.
Could joint marketing of Open Look have been Sun's only interest in the Amiga, spurring rumors about a licensing deal? It doesn't seem impossible to me that a number of Sun suits visited Commodore (or vice versa) to discuss this joint marketing, and that internal rumors then started spreading at Commodore: Something something marketing, something something Sun's Open Look, something something license.
Could Sun and Commodore have discussed a joint workstation venture years before the launch of the A3000UX, leading to rumors later on? If a deal between Sun and Commodore is to make any kind of sense, I believe it should have taken place some time during 1988, when Commodore announced the A2500UX. As a UNIX offering it could certainly have been competitively priced for the low-end market at this point in time, especially coupled with the A2024 monitor which was capable of displaying four grayscales in 1024x800. Alas, the Sun-Commodore saga hinges on the A3000UX which, in my mind, makes it rather unrealistic.
Finally
When doing research for this text, I unsuccessfully tried finding and contacting people who were more directly involved in Commodore's UNIX effort. If anyone out there sees this and would like to set the record straight, please don't hesitate to contact me.
Otherwise - bar a sudden discovery of official meeting notes or contract drafts - I guess we'll all have to make our own due diligence.
