I worked for Commodore too, 10 years later, In Holland, near Schiphol, we just agreed never to build any legacy systems anymore. Escom germany took them over, same company that did these UNIX systems. I did only PC related issues, selling them as Escom system, next to the Colani towers....
I don't have any details but the 900 was in the works before Jack Tramiel left Commodore in early 1984. I know this because he told me about it as did the folks at Mark Williams when Coherent was ported to the Atari ST.
The first version was definitely in the works before Jack left Commodore. As I understand it, there were three separate teams working on it at some point... they had problems. The final team, well after Jack was gone, was George Robbins and Bob Welland. The bit-blit chip was Bob's design, which was going to an add-on option to the base monochrome megapixel display. There was also going to be a multi-serial card to support multiple text terminals. As I understand it, after the Amiga purchase, Commodore management didn't feel they had the resources to launch and support to 16-bit machines at the same time. This was not a popular decision, particularly given that the market segments targeted for C900 and Amiga1000 couldn't be much more different. After the C900 was cancelled, George and Bob went on to create the Amiga 500. I was doing the new A2000, based off the A500 architecture, in concert in 1986, and Bob had already started working on a 32-bit expansion card that could run some version of UNIX. I designed better support for taking the alternate processor into the A2000's new Buster chip. Once the A2000 was entering production, I helped Bob with his 68020 add-in board. I needed to tweak it a bit to work properly in AmigaOS as well as booting to UNIX. Commodore had brought back some of the C900 software folks, Johann George and Rico Tudor, to work on "AMIX", the first version of UNIX for Amiga. Rico's own windowing system -- still the fastest I think I've ever used -- was part of that, apparently similar to what they had on the C900, though I had limited actual exposure to working C900s. AMIX was a bit tight in the 2MB of 32-bit RAM on the A2620... when I was offered the chance to make a 68030 version (after Bob had left for Apple), I was able to get 4MB in there, and an add-on connector for more DRAM.
I wonder the same thing too. Though I cant really see this machine really becoming all too popular. It would have needed to be a powerhouse to distinguish itself, and by this point 286 machines were also coming out which would have out performed the poor z800. There were low cost unix workstations from Digital with microvax and AT&T using 68k cpu's. Which also were more sophisticated and performant than the z800. Commodore would have had an edge in price for sure, but it would have still been far more expensive than budget unix PC's of the day. I cant help but think this machine would have sat in some sort of no mans land in the market.
This 900 was already obsolete before release. The death of the amiga really was sad, but it was inevitable when you take an entire industry of open PC platform parts makers and get them competing. Only once we get to the Apple 2018+ do we find one company with the trillion dollar valuations and assets needed to go vertical in the gigahertz clock and terabyte storage era of computer power.
Jack Tramiel could have kept commodore alive perhaps instead they dumped him and he used Atari to destroy commodore, sad story 😪 it made Microsoft happy Atari and Commodore destroying each other.
Atari too, really. Both of them made so many boneheaded errors including wasting endless time, money, and potential on interesting but rapidly abandoned vaporware whilst not advancing their actual products anywhere as much as they should have (and - particularly, as they're a 68000-based rival - Apple very much did). This thing sort of strikes me as a vaguely ST-like machine in some ways, but two years before the actual ST and probably with more development put into it pre-release. If they'd actually bothered to push the button it could have been massive ... and I bet a more home user based variant with a rationalised feature set and more colourful lower rez graphics and actual sound would have come along soon enough (and matching add-in cards for a "serious" one with a case better designed for expansion). With Commodore never bothering with Amiga after that, Shivji et al probably sticking around, and Atari picking up the miggy team without interference... Basically the history we know would have been kind of reversed...
@@WarrenPostma In the alternative timeline, the 900 would be released with the Amiga corp never purchased. Sales would have been very minor. The 8 bit line would likely have had more successor models to expand colors, speed, and resolution. Gradual sales decline in these legacy 8 bit lines and the lack of a true 16 bit system would cause Commodore to panic and pouring cash into some 68k based system designed internally. By then, it was too late, with the 8 bit sales stagnating and cash investment losses, Commodore gets bought out or insolvent by 1990/1991.
It would need interlacing to display 400 scan lines anyway... making it an RGBIi monitor. Either that or they use EGA with settings that are slightly out of spec.
That being said I think Commodore made some interlaced video cards, and Commodore monitors from this era (1983) were typically interlaced NTSC composite color video, the 8563 was not not CGA-compatible.
There were 400-line monitors at the time, generally running around 25kHz instead of the usual 15 (or the 18 of MDA, 22 of EGA). Tandy, Olivetti, and a few others used such, and often they were RGBI as well. They kept CGA compatibility with the PC ones fairly simply by line doubling in lower rez modes. I'd think it quite surprising to have something like this hooked up to a basic 15kHz screen, even if the card had its tech reused at lower rez in the C128. I wouldn't be super surprised however to find that the 1024x800 mode is interlaced, unless it's quite an advanced card. It's quite a demanding resolution for any hardware of the time, and you'd need to do that to get 800 lines out of, e.g. the Atari ST hi-rez mono monitor. Commodore wouldn't see that kind of rez again until the A2024, which had its own built in framebuffer (possibly something related to the card in this system?), running a rebadged 64kHz workstation monitor, and took video from the Amiga either in hi-rez lace mode and deinterlaced it for output, or as sequential lower resolution tiles and reconstructed them into a larger image. Though probably it wouldn't be the same 25kHz type even if it would have been tempting to do so... they already tended to only be around 50 to 60Hz, so interlace at that rate would be just as painful as on the Amiga with a regular CRT. Probably more like the 32kHz of VGA or 36kHz of the ST (so you were at least starting from 70Hz refresh) if not higher. And it wasn't at all unusual to have different dedicated monitors for different video modes at the time after all. And it doesn't LOOK interlace-flickery in the demo video here, but I dunno whether you'd see that under such filming conditions?
@@tahrey ב''ה, not sure why but I think the high res mode was actually NI on this one, or perhaps something like delaced for usable "professional" output without headaches. Just loosely related, the history of Commodore internally beginning to understand the importance of software compatibility in this era used to be interesting. Even without that this machine could have found a niche as a storage server for Amigas and would have moved CBM a bit ahead on networking that Apple and Atari ended up ahead on.
Yes - I was about to post the same thing - RGBI usually means RGB with Intensity - which is usually a 15 colour standard using TTL RGB to give you Black, White and 6 base primary and secondary colours (RGBCMY), but an additional TTL Intensity bit gives you high and low brightness options for White and RGBYCM (I don't think you get two blacks). The 400 lines - by that point - was more likely to be implemented by using a higher frequency line scan rather than interlacing (i.e. 400 progressive lines refreshed at frame rate, not 200 line fields refreshed at a field rate of twice frame rate)
I vividly remember being at College in 1986 and telling people that C= had planned a UNIX workstation, later when the 3000UX came out I thought I'd got it wrong. Finally, great vid.
Correction: Compute! Is a US not a Canadian mag 🇺🇸 Thank you so much for watching. If you enjoy my videos then you might like to become an Official Cave Dweller at patreon.com/rmcretro to support the channel, the museum and our projects. If you'd like to visit The Cave you can check out retrocollective.co.uk. Thanks so much! Neil
There is a Canadian magazine called The Transactor which is well worth reading. Quite technical but for the 8-bit Commodores. There was also an Amiga Transactor for short time.
Funny thing: I live in the town where the "Papst Motoren" company is situated. St. Georgen, Germany, Black Forest. 🙂 7742 is the old German-zip-code, that was used before the country switched to a 5-digits zip-code. It's quite a small town, too, with only about 13.000 inhabitants.
I remember Pabst fans being pretty popular in PCs 20 years ago or so. I think they stopped marketing directly and instead just operate as an OEM supplier now.
@@Nukle0n Yeah, I think they still produce fans, but they changed their target audience a bit, more towards industrial, I think. But they're still doing pretty well.
I remember Papst fans being "the quality ones" back in the day. They were expensive but they did last about forever. I actually still have some late 80's Papst fans which work just like new.
Commodore's marketing reminded me of of a joke ex-Commodorian Guy Wright used to tell: "Commodore couldn't effectively market a cure for death if they had one".
Hi, Italian here! I scrolled some comments, but I haven't found someone talking about this. I can confirm that that "Polloce" is a typo. It should be "Pollice", which is the translation for "Inch". "Polloce" is not a word in italian. The nearest word is "Pollo", which is indeed a chicken. *However, there's a thing that can be mildly interesting* : the italian text from 23:08 is weird to read for an italian (Just like my english, sorry 😅). It is somewhat correct, but it seems that the one that wrote that text (Maybe a translator) didn't have a lot of context. For example, the phrase "Contiene memoria addizionale di 128KB Bit-Mapped Display" doesn't make a lot of sense. A better example I can make is "Include 128KB di memoria addizionale per il display Bit-Map" (And I don't have the exact context too, it seems that the memory is inside the monitor). Other oddities I can find are - "UNIX-Compatibile", which should be "Compatibile con UNIX" - "Includendo", which should be something like "Ciò include" - "Finestre-testo [...] finestra", which should be "Finestre testuali, con la possibilità di ridimensionarne i caratteri per adattarli alla finestra" - "Finestre grafiche [...] scale", which should be "Finestre grafiche ridimensionabili con diversi valori di scala" or something similar I'm open for better translations or corrections and any help!
Serial terminals certainly are capable of high resolution graphics. Tektronix were the leading company making graphic terminals in the 70s, using a storage tube so that the terminals didn't have to contain enough ram to support their very high resolution of 1024x1024. Other manufacturers produced Tektronix compatible displays which used normal raster scan video with backing store, and in my first job in the 1980s I used some of these connected with serial ports to a prime minicomputer to do CAD. It worked pretty well. Also, RGBI means RGB plus intensity. I.e. they used an IBM CGA compatible monitor with that video board.
There are various videos here on UA-cam decorating the Tektronix terminals. Search for Tektronix 4054 to see what the later note sophisticated state tube terminals could do.
Yup & that's how they wasted time AND capital. A lot of it. Some of it stuck, such as the C-64, Amiga & perhaps their calculators, but the rest slipped off the wall. But in a way, they got into computers by accident in the first place, through the butterfly effect from typewriters & then into electronic calculators & then thinking about homebrew kits which turned into the KIM & then the PET. I did not like how they went down the road with the CDTV & CD32 either, but I guess everything critiqued comes from hindsight.
@@robwebnoid5763 In the book "On the Edge" it was said that Jack Tramiel would take prototypes to exhibitions like CES and see which ones generated the most interest. As for their start in computers it was to get revenge on Texas Instruments who almost forced them out of the calculator market. They succeeded. Back in the day I paid £59 for brand new TI99-4A computer in John Lewis department store while the C64 was selling well at three times the price.
@@MrDuncl ... Yes & that would not have all happened without Chuck Peddle. For me, another wasted effort is the 264/364/116/TED line. It would have been fine if they made 1 or 2 models, but they made too many different ones, trying to see which one would find engagement with the public. Just another one of those "which one might gain interest" as you said. And it was all to compete at the low end once again, with the likes of the T/S 1000. Little did Tramiel know that the T/S 1000 would have been sort of a bust anyway because the public did not want a computer that was so cheap that it did very little. I played with the T/S at department stores back then & it was cool although I already had my sights on the C-64 & am glad I did, 1st bought in 1983. I did see those Memotech add-ons for the T/S in magazines & that was somewhat of a temptation to put all that together to make a better T/S machine, but I think the C-64 trumped all of that anyway. They should have just dumped all of that 264 black-colored concept once Tramiel quit. But again, it's all hindsight. Half of this I already knew from reading over the decades but the other half is watching Dave Haynie & Bil Herd talk about this for the past decade or so. Fun stuff. I did also play around with the TI 99/4A in those same department stores. My first impression of it was not good as the floor demos had them attached to TV's, so the screen was awful, too much rainbowing & interference. Today, there are graphics mods for the TI to make the screen look cleaner & sharper, as also many types of mods for the C-64 & many other orphaned machines, including mods for the T/S & TRS Coco's. Imagine if all these mods happened back in the heyday of these machines, heh. I still have 3 C-64's, with peripherals, disks & paperwork (manuals, magazines, etc).
Same thing with Apple as it crept into the mid 90s until Jobs took over from the brink of it going insolvent. Actually, much worse, with its extremely bloated and ineffective, yet super money hemorrhaging, management style.
Now all you need to find is an Atari Transputer Workstation. Maybe this Commodore was canned because there was no signs of software support for it. At work in about 1990 they bought a load of HP Apollo Workstations to run Mentor Graphics CAD software only for Mentor to announce that future versions of their software would only run on Sun Workstations. The workstation market was already consolidating to just Sun and Silicon Graphics back then.
This was a wonderful time in computers! Commodore wasnt the only company coming out with computer gear. I did think it a bit crazy Commodore seemed to be doing so many different things! It was delightful shopping in stores and actually seeing much of this gear. Then there were the magazine ads and articles. I couldnt keep up.
Thanks Paul for your detailed reply, I did not know the history. At work in 1981, I got a new Dec PDP11-23i for my programming work and soon had MS Xenix installed on it by Logitech (they came out from their London office). Unfortunately this did not have enough memory to run the Vi editor (and using the ed line editor for C programming was painful). I then got a Unix Hot-Box with a Motorola 68000 that could run Vi well, but required a graphical terminal for my programming work. The graphical terminal gave slow graphics, so I ended up looking at Work Stations. The Apollo was too expensive, the Sun 2 was not easily available, hence I got my employer to buy an MG1. However, the software support was never really there, so after a couple of years I ended up with getting a Sun 2 workstation for my work. I kept on using Sun workstations happily for many years. Since about 2000, the same work could be done on any home PC.
It never occurred to me to ever hear about a new Commodore model. Until today. The front panel bears a distinct resemblance to the design of the A2000 front panel. Since this unit was also built in Braunschweig, this would also indicate that the 900 was destined for assembly in Germany. It is possible that the production line (tooling, plastic molds, etc.) for the 900 was already in place but then not used, so the A2000 design was derived directly from the 900. This approach would fit Commodore's habit of using whatever was already available to reduce costs.
The Z8000 was a rare chip indeed, considering the success of the Z80, it's surprising how little uptake it got. It's architecture used segmented address like the Intel chips, which was less pure than the 68000 linear addressing model.
I remember reading long ago that the Z8000 was perhaps the last non-microcoded complex CPU. This caused Zilog all kinds of difficulty when trying to develop the CPU, since every instruction is implemented with a bunch of complex logic instead of as a series of micro-instructions. This approach uses fewer transistors than microcoding, but it's much more complex to develop and debug. The initial runs of the Z8000 would have some bugs, and this caused delays in getting working parts to interested parties. Since it wasn't backwards compatible with anything, there was little reason to choose it over competitive CPUs.
That IEEE-488 port is otherwise known as GPIB or HPIB, which was a common interface used and still in use on all sorts of lab test equipment. It was also used for factory automation and for connecting pen plotters. Its very expensive to add that to a modern PC these days.
I always find it interesting how Commodore, a company sort of built from the ground up to be laser-focused on the US market, ended up being enormously successful everywhere *except* the US. I imagine there was a real need for affordable Unix workstations in Europe and this was Commodore's sort of half-hearted attempt to fill that need. Also I wonder if this had any influence on the ill-fated Amiga Unix offering that would come a few years later.
I had a friend with an Atari ST back in the late 80s/early 90s. He introduced me to mod tracking. I was overjoyed when I found a macintosh mod tracker, 'Meditor', which ran beautifully on my family's Mac IIsi. But I later got a 386 DX33, my first PC in 1995. Last year I released an album of the mods I made in Meditor... It's not good but feels good to have it out there for the world to see now!
This is a fantastic addition to the Commodore/Amiga collection in the Cave. I'd love to see follow up videos on the restoration of this unique piece of hardware.
I remember seeing this at newteks offices back in the 90s. I was in involved in teaching video toaster classes and we occasionally had rare or not tested machines. I do remember a 900. This has got to be the same unit. Either way very nice. Commodore always surprised me when I started to consider brand changing ....it just knew my feelings hahaha nice vid neil
@21:06 That could very well be the system oscillator. It's quite common to see them at twice that of the targeted frequency so it could very well be running at 6 MHz
A diamond in the rough that 900. To see it in this form truly does make one wonder what could have been under proper management (ie not commodore). I can only imagine how much money was spent getting to the prototype stage only to have it cancelled in the end. I shall sip a cup of tea in the 900’s honour with my RMC mug tonight.
I enjoyed your video, which was packed with useful information. I had been trying to find information about the Commodore 900, but my search was mostly fruitless - I only managed to find a small amount of information. However, your film contained a wealth of information on the subject, making it a very interesting watch.
Phenomenal video! I used to subscribe to Compute! and Compute's Gazette as a kid and I remember the hype about the Z Unix machine. Glad to see such a fine example in your possession Neil. Quite an interesting story.
Another very interesting video. The Zilog Z8001 CPU was sadly not implemented in many computers and besides this Commodore, #Olivetti was most likely the only manufacturer that actually produced a real computer with the Z8001 CPU - The Olivetti M20 - back in 1982. Keep up the amazing job telling us the history of all the interesting computers.
It also looks rather M20 / AT&T6300 ish to my eye. And it even sounds like it has the 25kHz video mode. Though wasn't it the M19 that had the Z8000, and the M20 was 8086 based with a Z8000 add-in card to run M19 software? (I was thinking it had some similarities before even getting to the comments fwiw)
Onyx Systems made a few models with the Z8001 having introduced one with the Z8002 and custom memory management hardware. These systems ran Version 7 Unix and later System III. I could easily imagine that people in Commodore saw these and saw an opportunity to make them cheaper, but the Z8000 was really the wrong horse to back, especially by 1984.
@@tahrey Just to clarify. The Olivetti M20 was the computer with the Zilog 8001 CPU. The AT&T6300 is the american version of the Olivetti M24 and that has the Intel 8086-2 running at 8 MHz ... You can see both of them explained on my channel
@@TiBosRetroComputers aha, thanks. I do tend to get similar model numbers mixed up. ....maybe a bit out of nowhere, but can you scope the H and V sync lines on the video to get the frequency for each of them by any chance? Been trying to flesh out the list of such things for old machines and the documentation for them is very sketchy and unreliable. Like you do the maths on the alleged CRTC register settings and get a completely nonsense result.
Tremendous work. Bravo! It seems a shame this design was cancelled when it was ready to go, but it is easy to forget just how crowded and diverse the field was at the time. Even a complete system represented a risk to the manufacturer.
I my self currently have 2 different Olivetti M20 computers both running. One of them even features the intel 8086 add on board extending the #Olivetti also to be able to run DOS
At uni on my A1200+Blizzard A1230+882+SCSI card, I ran Debian Linux, but that ran really hot and would just cut out when using it. It needed a bigger case, if only they'd have built tower cases for those machines and at a reasonable price, it wouldn't have been so bad.
Watching this reminded me that I have actually had a play with one of these machines at a Commodore conference that was launching the new updated business PC's (I worked for a Commodore business partner), it wasn't mentioned in the talk but was up and running with a couple of termnals.
Wow , amazing.....first time that I hear about this machine......one wonders how it could have been ....great video congrats..I hope the specialists can fix it, I'll like to see more of it 🙂
This is a great insight into a machine that deserved to be released to market at the time. I think it would have sold well. I used UNIX based machines in college in 88 and 89 and I found it to be very usable and extremely stable. So UNIX was a good move I think for the time. What could have been, eh? . Such a great shame. But this technibble provides a lot of great information which is excellent to learn about this machine and also to actually see an example. Thanks Neil..
Nice work and super interesting machine! Too bad it got scrapped. I can confirm that the keyboard is a standard Commodore PC keyboard, it's the exact same model as the one that originally came with my Commodore PC 10-II, also with the German layout.
I would love for you to do a video on the Amiga Transputer which Commodore showed off at the NY Commodore show in the late 1980's. It looked really interesting, but it was never produced and marketed.
@@teekay_1 Well, if by "They" you mean Tim King of Metacomco, maybe. The "Commodore Transputer Project" was simply a board designed for the Amiga 2000, at our engineering office in Braunschweig, Germany, some years after Jack Tramiel and his associates had left the company. Tim was working on a new company (Perihelion Software) and operating system (HeliOS) for the Transputer and apparently evangelizing it around the industry. Atari's Abaq was a complete packaged system internally based on a modified ST plus a Transputer module, but far as I know there was no connection to the project done at Commodore.
@@DaveHaynie Interesting Dave. I'd read somewhere differently. But you were there so you know what actually transpired. I did see a demo in NY I believe sometime in the late 80's at an Amiga Expo. I thought it was pretty cool.
Excellent video, I have never heard of the Commodore 900 before! This is one fascinating machine, I really hope you can get it up and running as curious to see how it looks and what it can do. Next question is software for it ...
I remember seeing the 900 in the `85 Commodore magazine. It was a few years later when I owned an Amiga 2000, I kept wondering why I had felt I had seen this thing before. Then I later found my copy of the mag and had that ah-ha moment. Also… It is said that the Atari ST is the true legacy of this machine. I've read accounts that the engineers that followed Jack Tramel over to Atari, brought the 900 plans over with them. Z8000 replaced with MC68000, CPM/68k instead of Coherent.
Is it, though? The ST used a very different set of off the shelf chips, as well as the different CPU. It’s always felt to me like the STs were an attempt to produce a cost-reduced version of the Mindset PC, but with a 68000 and Mac-like OS. (There was also talk at the time of Atari going with National Semiconductor’s NS32000 chip, but it was ultimately determined the 68000 was faster.)
Yeah, I was thinking the same all the way through the video, from fairly early on. I even had a bet that there would be a 640x400 video option and let out an audible "ah-ha!" when that was revealed... Plus there's the blitter, daisychain peripheral interface, 512k (or 128k / 2M) memory bank size, etc, though it took a while before they made a pizzabox machine, and nothing properly Unix compatible or with Workstation grade hi-rez until the TT in like 1989. I do now need to go look up the one low quality photograph I found of the vaporware "EST" from like '86-87 that would have had slightly better video and an 68020, so may have had a better chance. I have a feeling that the case looked VERY similar to this... Plus of course Shivji was involved in both, and the ST did get whizzed up remarkably fast ... with a bit of a make-do soundchip. And a VT52 terminal emulator as a default pack in utility, and a serial based debug / diagnostic feature... Even the first OEM hard drive was a 20MB ST225 based item, though an external one. Though I guess the original floppy was only 360k instead of 1.2M, before going up to 720. And there was the odd throwing-in of cartridge, MIDI and mouse... and the shared video memory instead of a separate card... Whether you'd even need full plans, because it's not the most complicated block diagram to remember, though? Just having been involved with the project would probably give you enough to go on, if you were in the position of heading up a development team at a major computer manufacturer. Maybe some of the low level stuff involved with actually making the raw low level electrical stuff work with the relatively wide and high speed bus etc might be valuable, but the general setup is fairly straightforward. More than a bit of the ST itself can probably be traced to a Motorola reference design for an S100 system, just with the video taking advantage of the higher speed memory that became available, so the C900 parts may be more conceptual than specific. Plans for the blitter would have been valuable however... just it took a little time to get it to a suitable state for manufacture then inclusion in the Mega and STe. And maybe there's some shared DNA between parts of the C900 video systems and the ST Shifter? And then into the TT as well? (clock multiplication and specialist high speed line drivers for extra high rez video, with the advantage of an 8 rather than 6MHz basis, so you can squeeze out some extra pixels for 1280x960 instead of 1024x800...)
@@sunspot42 Maybe they gave those other chips a look-in because the Z8000 had proven lacklustre in testing and they didn't fancy repeating the same mistake? And the 68000 had already proven itself in the Lisa, which apparently was used as a dev tool for the ST, so it was kind of expedient, as well as having good reference designs and support chip families available. The Z8000 really is a bit of a weird choice, especially for either of those two companies which hadn't done anything with Zilog (or the Intel "inspirations") before, and instead used opposite-way-up 6502s (...derived from the 6800 of course, which the 68000 retained some support / peripheral support with). The 68k really was a much more logical progression. Plus the advantage of not having segmented memory, running at the rather more optimal-for-the-time 8MHz without needing to pay for and waste a bit of something binned for 10, etc. But, there were a lot of similar designs around at the time, if you were trying to make something affordable with meaningful 16-bit power at the time, your choices for processor and system design were kind of limited. Even those which used alternative CPUs (like the Archimedes) ended up bearing an uncanny resemblance. The comparisons are inevitable. It's just that given the personnel and other items of heritage, plus certain features of the design, made me sit up and go "hang on a minute" even not very far into the video. The ST sort of seemed like a bit of a true C64 / C128 successor if anything (vs the Amiga being seemingly more related to the Atari 8-bits), but lacking a little on the audio and hardware video acceleration (and lack of full ANSI 80-column colour) front. With the C900, a few more pieces fall into place. There's bits of each in there, conceptually.
I used to own a Z8001 Zilog Unix (ZEUS) machine, great looking toy, but even at the time was somewhat limited. The speed was not great and the limited RAM made it pretty useless for anything other than small compiles and text editing. I think the cost of enough resources to make Unix usable was the real reason why a lot of machines like this got scrapped in development, it would be quite some time before the capacity and cost of RAM made (even a limited 16 bit) Unix worthwhile.
Compute! was a US based publishing company. They published computer books and magazines. ABC Publishing (America Broadcasting Company) was their parent company.
Depending on price and peripherals, this could have replaced mini computers, the mini I was using was configured in a similar way, but with much less memory & only 5m of hard drive space. So I could see this as upgrading small businesses at the time. Could have been interesting.
I had no idea about the 900 or the UNIX variants of the 2500 and 3000. I had both the 2500 and the 3000 which did have the 68XXX memory management chips, but Commodore never did anything with segmenting up memory. Instead, all of the apps could write all over each other.
The MMU in the A2500/20 and A2620 was entirely there to support UNIX. The later machines got it for free with the 68030 and 68040 processors. The time wasn't right for a protected AmigaOS, based on the need to support lower-end systems, which were the majority of the Amiga computers Commodore sold. However, the software team and technical support group released various MMU-based development tools (Enforcer, MemMunge, etc) that worked to help find bugs that could cause catastrophic failures. This was also encouraged enough that code reviewers caught on and generally ran these tests for reviews... you didn't want a reviewer to report Enforcer hits with your new software!
Super-neat. I love obscure and never-quite machines like this. Given that the Z8001 never really went anywhere, I wonder if the lack of adoption generally was part of the decision - Zilog might've ended up having a chip widely used by only one maker. That's in a bad place to be, if you're talking about moving forward with an architecture, something which probably wasn't real obvious in 1982 or 1983, but was starting to become quite clear in 1985.
About the Italian translation at 17:09. The word polloci it's wrong, the right one is pollici (as also suggested by the translator program). I presume a typing error during the input for translation.
Thanks, I'd never heard of this machine before and it's really interesting! BTW what happened to part 2 of the Enterprise computer video? Part 1 was over a year ago now.
Compute! Magazine was headquartered in Greensboro, NC (the "Canada" reference on the magazine cover is the Canadian dollar price). The domestic magazine market for the USA actually includes Canada and the US, as does the domestic box-office/movie market.
There are quite some informations about this machine in Brian Bagnall‘s Commodore book („Commodore - A Company on the Edge“, in Germany the book is sold under the name „Volkscomputer“. See page 258ff in the german edition about the evolution and canceling of the machine.
I remember the VIC-20 (aka VC-20) being marketed as the "Volkscomputer" in Germany, The Netherlands and Belgium. I also remember waiting for home computers with the Z8000 and Z80000 "mainframe on a chip" CPU's to become available during the early to mid-80's and killing the inferior 8088 IBM PC (which I considered to be "too little and too late" - boy, was I wrong!), but this never happened.
@@koenlefever The "VC" came of course since "VIC" spelled out in German sounds just like the word for "fuck". So they changed it to VC. Ret-conning it into an abbreviation for "Volkscomputer" was probably just an idea by the marketing guys...
It would be interesting to see if that keyboard is pin-compatible with the Amstrad PC 1512/1640. The layout looks exactly the same and it has the same Lamps-under-keys scheme for Caps/Numlock
The 68k architecture is 32-bit, as the instructions and registers were 32-bit. The 68000's implementation relied on hardware which could only operate on 16 bits at a time, so it came with a performance penalty, but the fact remains that as far as software was concerned, it was no different than a 32-bit clean CPU.
@@Henrik_Holst The 24-bit address allowed for some bad programming habits. For example, Microsoft BASIC was using those unneeded top 8 address bits for some kind of extra data fields, so it failed hard once we had machines out with memory beyond that first 24-bits which, of course, required 32-bit clean code. Once we had machines with MMUs, developers could run Commodore's Enforcer system and test all this even without having that extra memory.
@@DaveHaynie of course people would abuse that :). The same happens today with 64-bit addressing since no one imagines that we will ever see such a large system. If I recall correctly e.g Intel uses up to 15-bits for pointer tagging. Everything repeats :), and oh btw huge thanks for your work on the Amiga that and the c64 was what teached me programming back in the day (more or less was forced to on the c64 since I didn't had either the datasette or a floppy initially).
no, at least not binary-compatible. the Z8000 supported equivalent instructions (it could do the same things) with equivalent memory addressing (it could get/put things the same way) but the two were not directly compatible. working from memory, a source-code translator was available that would map the Z80 instructions to Z8000 instructions, which could be assembled and linked to create an equivalent (but different) executable program that behaved the same way as the origina Z80 program.
It shows how little room is needed for text and numbers. When pictures and videos cropped up, then we needed way more RAM (and larger removable disks too e.g. SyQuest, IoMega). As nice as the C900 might have been, my friend and I just ran Xenix on Compaq Deskpro 386s and added multiport serial cards. Maybe Commodore realized that it wasn't worth competing against such commodity "micro mainframes".
Those of us programmers that learned to code on a Vic-20 with 3.5KB of RAM learned to optimize code/data/etc. Sadly, this is mostly a lost art now and hence the bloat of most software.
@@GeoffSeeley very true but you trade that for APIs that allow software to work across a huge variation in hardware. Most modern software could probably run on a Core 2 Duo 4GB system in a Windows 7 PC and on an AMD 64 Core Threadripper 64GB system in a Windows 11 PC. Back then you had to tailor your software very specifically for each hardware. Edit I'd say the changes are better for developers to actually make money, as software gets bigger and bigger, being able to sell to more people gets more important.
No, the lack of memory always made the systems primitive. The text-only interfaces, hardware monospace fonts, single-byte character sets, US-centered software...
Music in this video is: 1AM OMW - Ballpoint All Parts Equal - Airae High Octane - Hampus Naeselius Innovations - From Now On Let There Be Rain - Silver Maple Lofty Palms - Yi Nantiro Message Between Worlds - William Benckert Revitalize - Yonder Dale The Farmhouse - Silver Maple
@@treychatillon6994 Makes me appreciative how much effort is put into these videos when he uses 9 different tunes for a 30min video on a niche subject. He had to sample multiple music tracks to find the ones he wanted to use, or at least have already sampled them to have in his library of acceptable music to use.
I had a Z8001 machine back in the early 80's from Whitechapel Workstations. This was a British competitor to SUN 2 workstations and similar Unix workstations of the time. I visited the company base in Whitechapel London for demonstrations a few times. The machine had a high definition black and white monitor for CAD and similar applications. They worked quite well, but it was very difficult for the company to compete against Sun Micro Systems and the Motorola 68000 ubiquity in the CAD market. I seem to remember some hardware/firmware limitations/bugs in the Z8001 chips in the virtual memory area. This was not helped either by Zilog as a company not being stable. I don't think the Zilog 8000 processors ever got much of a market. Perhaps Commodore hit the same issues.
Whitechapel's MG-1 was a 32016-based workstation, and the NatSemi chipset had quite a few problems that inhibited adoption of the 32000 series. Indeed, there are signs in reviews of the MG-1 that WCW might have had to use some workarounds for problems with the memory management unit. Acorn Computers were supposed to deliver Unix on the 32016 but never did, with the MMU also featuring as an obstacle in their plans, although one can argue that Acorn didn't seriously pursue that product and didn't commit the resources to it, either, going through near-bankruptcy at that time.
@@Ray-ds5dc No worries! Great to hear that you used an MG-1 back in the day, though. I hear it was briefly popular in London financial institutions. As for Z8000-based machines, reading up on them a bit, I think that one of the first ones to reach the UK was the Onyx C8002. That was actually based on the Z8002 with custom memory management hardware because Zilog's MMU appears to have been delayed, this pushing back the introduction of Z8001 machines. The C8002 seems to have been competing with PDP-11 machines, which would not have been a viable strategy for very long, however. So, there are definitely parallels between Zilog and NatSemi in terms of problems delivering their chipsets. Interestingly, vendors shipping 68000-based systems also developed their own memory management hardware until the 68010 came out. At that point, the Z8000 would not have been very attractive, and by 1984, when the 68020 came out, it seems that various companion chips for the Z8000 were still not available. There would have been many reasons at that stage for anyone using the Z8000 to jump ship to something else.
... Remember ... once upon a time there was a multi user multitasking operating system written in BCPL called TRIPOS ,, from Cambridge University ... it was ported to the 68000 and used in a machine called the ... Amiga .... so,, a business system using the early version of Amiga OS & serial terminals, was already in operation ...
I worked at a Commodore authorized dealer as a repair technician in the late 80s and this is the first time I've ever heard of this machine.
I worked for Commodore too, 10 years later, In Holland, near Schiphol, we just agreed never to build any legacy systems anymore.
Escom germany took them over, same company that did these UNIX systems.
I did only PC related issues, selling them as Escom system, next to the Colani towers....
I don't have any details but the 900 was in the works before Jack Tramiel left Commodore in early 1984. I know this because he told me about it as did the folks at Mark Williams when Coherent was ported to the Atari ST.
Hi Leonard, thanks for commenting, would you be interested in an interview some time? I’m sure a lot of our viewers would love to hear your memories
Wait, what? Well, that's another glaring link isn't it...
The first version was definitely in the works before Jack left Commodore. As I understand it, there were three separate teams working on it at some point... they had problems. The final team, well after Jack was gone, was George Robbins and Bob Welland. The bit-blit chip was Bob's design, which was going to an add-on option to the base monochrome megapixel display. There was also going to be a multi-serial card to support multiple text terminals.
As I understand it, after the Amiga purchase, Commodore management didn't feel they had the resources to launch and support to 16-bit machines at the same time. This was not a popular decision, particularly given that the market segments targeted for C900 and Amiga1000 couldn't be much more different. After the C900 was cancelled, George and Bob went on to create the Amiga 500. I was doing the new A2000, based off the A500 architecture, in concert in 1986, and Bob had already started working on a 32-bit expansion card that could run some version of UNIX. I designed better support for taking the alternate processor into the A2000's new Buster chip.
Once the A2000 was entering production, I helped Bob with his 68020 add-in board. I needed to tweak it a bit to work properly in AmigaOS as well as booting to UNIX. Commodore had brought back some of the C900 software folks, Johann George and Rico Tudor, to work on "AMIX", the first version of UNIX for Amiga. Rico's own windowing system -- still the fastest I think I've ever used -- was part of that, apparently similar to what they had on the C900, though I had limited actual exposure to working C900s. AMIX was a bit tight in the 2MB of 32-bit RAM on the A2620... when I was offered the chance to make a 68030 version (after Bob had left for Apple), I was able to get 4MB in there, and an add-on connector for more DRAM.
It's always a little sad to think of what Commodore could have been with proper management.
I wonder the same thing too. Though I cant really see this machine really becoming all too popular. It would have needed to be a powerhouse to distinguish itself, and by this point 286 machines were also coming out which would have out performed the poor z800. There were low cost unix workstations from Digital with microvax and AT&T using 68k cpu's. Which also were more sophisticated and performant than the z800.
Commodore would have had an edge in price for sure, but it would have still been far more expensive than budget unix PC's of the day. I cant help but think this machine would have sat in some sort of no mans land in the market.
This 900 was already obsolete before release.
The death of the amiga really was sad, but it was inevitable when you take an entire industry of open PC platform parts makers and get them competing.
Only once we get to the Apple 2018+ do we find one company with the trillion dollar valuations and assets needed to go vertical in the gigahertz clock and terabyte storage era of computer power.
Jack Tramiel could have kept commodore alive perhaps instead they dumped him and he used Atari to destroy commodore, sad story 😪 it made Microsoft happy Atari and Commodore destroying each other.
Atari too, really. Both of them made so many boneheaded errors including wasting endless time, money, and potential on interesting but rapidly abandoned vaporware whilst not advancing their actual products anywhere as much as they should have (and - particularly, as they're a 68000-based rival - Apple very much did).
This thing sort of strikes me as a vaguely ST-like machine in some ways, but two years before the actual ST and probably with more development put into it pre-release. If they'd actually bothered to push the button it could have been massive ... and I bet a more home user based variant with a rationalised feature set and more colourful lower rez graphics and actual sound would have come along soon enough (and matching add-in cards for a "serious" one with a case better designed for expansion). With Commodore never bothering with Amiga after that, Shivji et al probably sticking around, and Atari picking up the miggy team without interference... Basically the history we know would have been kind of reversed...
@@WarrenPostma In the alternative timeline, the 900 would be released with the Amiga corp never purchased. Sales would have been very minor. The 8 bit line would likely have had more successor models to expand colors, speed, and resolution. Gradual sales decline in these legacy 8 bit lines and the lack of a true 16 bit system would cause Commodore to panic and pouring cash into some 68k based system designed internally. By then, it was too late, with the 8 bit sales stagnating and cash investment losses, Commodore gets bought out or insolvent by 1990/1991.
RGBI doesn't mean "RGB interlaced", but "RGB with intensity" - basically a CGA-compatible monitor port.
It would need interlacing to display 400 scan lines anyway... making it an RGBIi monitor. Either that or they use EGA with settings that are slightly out of spec.
That being said I think Commodore made some interlaced video cards, and Commodore monitors from this era (1983) were typically interlaced NTSC composite color video, the 8563 was not not CGA-compatible.
There were 400-line monitors at the time, generally running around 25kHz instead of the usual 15 (or the 18 of MDA, 22 of EGA). Tandy, Olivetti, and a few others used such, and often they were RGBI as well. They kept CGA compatibility with the PC ones fairly simply by line doubling in lower rez modes. I'd think it quite surprising to have something like this hooked up to a basic 15kHz screen, even if the card had its tech reused at lower rez in the C128.
I wouldn't be super surprised however to find that the 1024x800 mode is interlaced, unless it's quite an advanced card. It's quite a demanding resolution for any hardware of the time, and you'd need to do that to get 800 lines out of, e.g. the Atari ST hi-rez mono monitor. Commodore wouldn't see that kind of rez again until the A2024, which had its own built in framebuffer (possibly something related to the card in this system?), running a rebadged 64kHz workstation monitor, and took video from the Amiga either in hi-rez lace mode and deinterlaced it for output, or as sequential lower resolution tiles and reconstructed them into a larger image.
Though probably it wouldn't be the same 25kHz type even if it would have been tempting to do so... they already tended to only be around 50 to 60Hz, so interlace at that rate would be just as painful as on the Amiga with a regular CRT. Probably more like the 32kHz of VGA or 36kHz of the ST (so you were at least starting from 70Hz refresh) if not higher. And it wasn't at all unusual to have different dedicated monitors for different video modes at the time after all.
And it doesn't LOOK interlace-flickery in the demo video here, but I dunno whether you'd see that under such filming conditions?
@@tahrey ב''ה, not sure why but I think the high res mode was actually NI on this one, or perhaps something like delaced for usable "professional" output without headaches.
Just loosely related, the history of Commodore internally beginning to understand the importance of software compatibility in this era used to be interesting. Even without that this machine could have found a niche as a storage server for Amigas and would have moved CBM a bit ahead on networking that Apple and Atari ended up ahead on.
Yes - I was about to post the same thing - RGBI usually means RGB with Intensity - which is usually a 15 colour standard using TTL RGB to give you Black, White and 6 base primary and secondary colours (RGBCMY), but an additional TTL Intensity bit gives you high and low brightness options for White and RGBYCM (I don't think you get two blacks). The 400 lines - by that point - was more likely to be implemented by using a higher frequency line scan rather than interlacing (i.e. 400 progressive lines refreshed at frame rate, not 200 line fields refreshed at a field rate of twice frame rate)
I vividly remember being at College in 1986 and telling people that C= had planned a UNIX workstation, later when the 3000UX came out I thought I'd got it wrong.
Finally, great vid.
I love deep dives into prototype PCs like this, especially Commodore since it was a huge part of my childhood. 🙏
Yes! Prototype PCs are, in my thinking, some of the most interesting machines possible. They're snapshots of what might have been.
Correction: Compute! Is a US not a Canadian mag 🇺🇸
Thank you so much for watching. If you enjoy my videos then you might like to become an Official Cave Dweller at patreon.com/rmcretro to support the channel, the museum and our projects.
If you'd like to visit The Cave you can check out retrocollective.co.uk.
Thanks so much!
Neil
That Keyboard is Swedish. And because we have ÅÄÖ (3 extra letters), the keyboards get another configuration than the standard "English"
There is a Canadian magazine called The Transactor which is well worth reading. Quite technical but for the 8-bit Commodores. There was also an Amiga Transactor for short time.
Funny thing: I live in the town where the "Papst Motoren" company is situated. St. Georgen, Germany, Black Forest. 🙂 7742 is the old German-zip-code, that was used before the country switched to a 5-digits zip-code. It's quite a small town, too, with only about 13.000 inhabitants.
I remember Pabst fans being pretty popular in PCs 20 years ago or so. I think they stopped marketing directly and instead just operate as an OEM supplier now.
@@Nukle0n Yeah, I think they still produce fans, but they changed their target audience a bit, more towards industrial, I think. But they're still doing pretty well.
I remember Papst fans being "the quality ones" back in the day. They were expensive but they did last about forever. I actually still have some late 80's Papst fans which work just like new.
"Zip code" is an American term. The general term in English is "postcode".
@@talideon Thanks. I wasn't aware of that, English not being my native tongue.
What a fantastic video. Such detailed information, well presented, and no fluff. Hats off!
This thing could have run an awesome multi-node BBS.
Wow. Amazing. I've been a Commodore enthusiast since the 1980s and I have NEVER heard of this machine. Thanks very much for making this video.
A genuine rarity! What a find. Hopefully we can see it working in a future video.
Definitely worth restoring. Would love to see how it works. 🤓
Commodore's marketing reminded me of of a joke ex-Commodorian Guy Wright used to tell: "Commodore couldn't effectively market a cure for death if they had one".
ב''ה, Warm Dead Bird
@@josephkanowitz6875Fantastic stuff, especially when covered in greasy wheat.
IIRC, Jerry Pournelle wrote a column in Byte magazine in which he praised the Amiga but lamented that it was "marketed like anthrax."
Hi, Italian here!
I scrolled some comments, but I haven't found someone talking about this.
I can confirm that that "Polloce" is a typo. It should be "Pollice", which is the translation for "Inch". "Polloce" is not a word in italian. The nearest word is "Pollo", which is indeed a chicken.
*However, there's a thing that can be mildly interesting* : the italian text from 23:08 is weird to read for an italian (Just like my english, sorry 😅).
It is somewhat correct, but it seems that the one that wrote that text (Maybe a translator) didn't have a lot of context.
For example, the phrase "Contiene memoria addizionale di 128KB Bit-Mapped Display" doesn't make a lot of sense. A better example I can make is "Include 128KB di memoria addizionale per il display Bit-Map" (And I don't have the exact context too, it seems that the memory is inside the monitor).
Other oddities I can find are
- "UNIX-Compatibile", which should be "Compatibile con UNIX"
- "Includendo", which should be something like "Ciò include"
- "Finestre-testo [...] finestra", which should be "Finestre testuali, con la possibilità di ridimensionarne i caratteri per adattarli alla finestra"
- "Finestre grafiche [...] scale", which should be "Finestre grafiche ridimensionabili con diversi valori di scala" or something similar
I'm open for better translations or corrections and any help!
Coherent is so fascinating. Never realized it was running on a Commodore machine!
Serial terminals certainly are capable of high resolution graphics. Tektronix were the leading company making graphic terminals in the 70s, using a storage tube so that the terminals didn't have to contain enough ram to support their very high resolution of 1024x1024.
Other manufacturers produced Tektronix compatible displays which used normal raster scan video with backing store, and in my first job in the 1980s I used some of these connected with serial ports to a prime minicomputer to do CAD.
It worked pretty well.
Also, RGBI means RGB plus intensity. I.e. they used an IBM CGA compatible monitor with that video board.
Thanks David, a very useful contribution
There are various videos here on UA-cam decorating the Tektronix terminals. Search for Tektronix 4054 to see what the later note sophisticated state tube terminals could do.
As an Amigan, I think I know a couple things about Commodore, yet I have never heard of this machine!
Commodore struck me as a company that just threw lots of different things at the wall to see what might stick
That's literally how they operated
Yup & that's how they wasted time AND capital. A lot of it. Some of it stuck, such as the C-64, Amiga & perhaps their calculators, but the rest slipped off the wall. But in a way, they got into computers by accident in the first place, through the butterfly effect from typewriters & then into electronic calculators & then thinking about homebrew kits which turned into the KIM & then the PET. I did not like how they went down the road with the CDTV & CD32 either, but I guess everything critiqued comes from hindsight.
@@robwebnoid5763 In the book "On the Edge" it was said that Jack Tramiel would take prototypes to exhibitions like CES and see which ones generated the most interest. As for their start in computers it was to get revenge on Texas Instruments who almost forced them out of the calculator market. They succeeded. Back in the day I paid £59 for brand new TI99-4A computer in John Lewis department store while the C64 was selling well at three times the price.
@@MrDuncl ... Yes & that would not have all happened without Chuck Peddle. For me, another wasted effort is the 264/364/116/TED line. It would have been fine if they made 1 or 2 models, but they made too many different ones, trying to see which one would find engagement with the public. Just another one of those "which one might gain interest" as you said. And it was all to compete at the low end once again, with the likes of the T/S 1000. Little did Tramiel know that the T/S 1000 would have been sort of a bust anyway because the public did not want a computer that was so cheap that it did very little. I played with the T/S at department stores back then & it was cool although I already had my sights on the C-64 & am glad I did, 1st bought in 1983. I did see those Memotech add-ons for the T/S in magazines & that was somewhat of a temptation to put all that together to make a better T/S machine, but I think the C-64 trumped all of that anyway. They should have just dumped all of that 264 black-colored concept once Tramiel quit. But again, it's all hindsight. Half of this I already knew from reading over the decades but the other half is watching Dave Haynie & Bil Herd talk about this for the past decade or so. Fun stuff. I did also play around with the TI 99/4A in those same department stores. My first impression of it was not good as the floor demos had them attached to TV's, so the screen was awful, too much rainbowing & interference. Today, there are graphics mods for the TI to make the screen look cleaner & sharper, as also many types of mods for the C-64 & many other orphaned machines, including mods for the T/S & TRS Coco's. Imagine if all these mods happened back in the heyday of these machines, heh. I still have 3 C-64's, with peripherals, disks & paperwork (manuals, magazines, etc).
Same thing with Apple as it crept into the mid 90s until Jobs took over from the brink of it going insolvent. Actually, much worse, with its extremely bloated and ineffective, yet super money hemorrhaging, management style.
Now all you need to find is an Atari Transputer Workstation.
Maybe this Commodore was canned because there was no signs of software support for it.
At work in about 1990 they bought a load of HP Apollo Workstations to run Mentor Graphics CAD software only for Mentor to announce that future versions of their software would only run on Sun Workstations. The workstation market was already consolidating to just Sun and Silicon Graphics back then.
This was a wonderful time in computers! Commodore wasnt the only company coming out with computer gear.
I did think it a bit crazy Commodore seemed to be doing so many different things! It was delightful shopping in stores and actually seeing much of this gear. Then there were the magazine ads and articles. I couldnt keep up.
Thanks Paul for your detailed reply, I did not know the history. At work in 1981, I got a new Dec PDP11-23i for my programming work and soon had MS Xenix installed on it by Logitech (they came out from their London office). Unfortunately this did not have enough memory to run the Vi editor (and using the ed line editor for C programming was painful). I then got a Unix Hot-Box with a Motorola 68000 that could run Vi well, but required a graphical terminal for my programming work. The graphical terminal gave slow graphics, so I ended up looking at Work Stations. The Apollo was too expensive, the Sun 2 was not easily available, hence I got my employer to buy an MG1. However, the software support was never really there, so after a couple of years I ended up with getting a Sun 2 workstation for my work. I kept on using Sun workstations happily for many years. Since about 2000, the same work could be done on any home PC.
You can see where the A2000/A1500 design got the inspiration from!
All systems did that design, the PC's too
This is literally a boring beige desktop exactly like 1000 earlier beige desktops from IBM, EPSON, and COMPAQ.
Yes you are right.
It never occurred to me to ever hear about a new Commodore model. Until today. The front panel bears a distinct resemblance to the design of the A2000 front panel. Since this unit was also built in Braunschweig, this would also indicate that the 900 was destined for assembly in Germany. It is possible that the production line (tooling, plastic molds, etc.) for the 900 was already in place but then not used, so the A2000 design was derived directly from the 900. This approach would fit Commodore's habit of using whatever was already available to reduce costs.
The Z8000 was a rare chip indeed, considering the success of the Z80, it's surprising how little uptake it got. It's architecture used segmented address like the Intel chips, which was less pure than the 68000 linear addressing model.
I remember reading long ago that the Z8000 was perhaps the last non-microcoded complex CPU. This caused Zilog all kinds of difficulty when trying to develop the CPU, since every instruction is implemented with a bunch of complex logic instead of as a series of micro-instructions. This approach uses fewer transistors than microcoding, but it's much more complex to develop and debug. The initial runs of the Z8000 would have some bugs, and this caused delays in getting working parts to interested parties. Since it wasn't backwards compatible with anything, there was little reason to choose it over competitive CPUs.
What a well researched video! Must have taken a lot of hours to get all the information together! Very impressive! Well done, Neil!
Thank you for all the videos you make. Always a great presentation of the best stuff from the past. Thank you again and keep it up!
That IEEE-488 port is otherwise known as GPIB or HPIB, which was a common interface used and still in use on all sorts of lab test equipment. It was also used for factory automation and for connecting pen plotters. Its very expensive to add that to a modern PC these days.
I always find it interesting how Commodore, a company sort of built from the ground up to be laser-focused on the US market, ended up being enormously successful everywhere *except* the US. I imagine there was a real need for affordable Unix workstations in Europe and this was Commodore's sort of half-hearted attempt to fill that need.
Also I wonder if this had any influence on the ill-fated Amiga Unix offering that would come a few years later.
I had a friend with an Atari ST back in the late 80s/early 90s. He introduced me to mod tracking. I was overjoyed when I found a macintosh mod tracker, 'Meditor', which ran beautifully on my family's Mac IIsi. But I later got a 386 DX33, my first PC in 1995. Last year I released an album of the mods I made in Meditor... It's not good but feels good to have it out there for the world to see now!
This is a fantastic addition to the Commodore/Amiga collection in the Cave. I'd love to see follow up videos on the restoration of this unique piece of hardware.
I saw ome of these at the June '85 CES. It was running a windowing demo rather than the "attract mode" demo shown here.
I was googling this yesterday, and I thought "Damn, YT is really on that targeted ish today."
Nope, its a new RMC!
Magnificent work! Thank you for sharing this rare gem and making an important bit of 16-bit Commodore history better known.
I remember seeing this at newteks offices back in the 90s. I was in involved in teaching video toaster classes and we occasionally had rare or not tested machines. I do remember a 900. This has got to be the same unit. Either way very nice. Commodore always surprised me when I started to consider brand changing
....it just knew my feelings hahaha nice vid neil
What a great video! I'm still learning so much about Commodore as a company all these years later and it's absolutely fascinating.
Wow, imagine if you could buy a Unix workstation in 1984 at Commodore prices..... this was such a missed opportunity.
Excellent video as always, had no idea about these.
Cheers Steve!
The "high capacity chickens" thing reminds me of when my school friend would jokingly refer to the "bumpy chickens" (chicanes) in SuperCars. :D
@21:06 That could very well be the system oscillator. It's quite common to see them at twice that of the targeted frequency so it could very well be running at 6 MHz
No-one does these style videos better than Neil. Excellent as always.
A diamond in the rough that 900. To see it in this form truly does make one wonder what could have been under proper management (ie not commodore). I can only imagine how much money was spent getting to the prototype stage only to have it cancelled in the end.
I shall sip a cup of tea in the 900’s honour with my RMC mug tonight.
I enjoyed your video, which was packed with useful information. I had been trying to find information about the Commodore 900, but my search was mostly fruitless - I only managed to find a small amount of information. However, your film contained a wealth of information on the subject, making it a very interesting watch.
Dude, that is sick! I'm so glad that these forgotten pieces of history have been found.
Phenomenal video! I used to subscribe to Compute! and Compute's Gazette as a kid and I remember the hype about the Z Unix machine. Glad to see such a fine example in your possession Neil. Quite an interesting story.
Another very interesting video.
The Zilog Z8001 CPU was sadly not implemented in many computers and besides this Commodore, #Olivetti was most likely the only manufacturer that actually produced a real computer with the Z8001 CPU - The Olivetti M20 - back in 1982.
Keep up the amazing job telling us the history of all the interesting computers.
It also looks rather M20 / AT&T6300 ish to my eye. And it even sounds like it has the 25kHz video mode.
Though wasn't it the M19 that had the Z8000, and the M20 was 8086 based with a Z8000 add-in card to run M19 software?
(I was thinking it had some similarities before even getting to the comments fwiw)
Onyx Systems made a few models with the Z8001 having introduced one with the Z8002 and custom memory management hardware. These systems ran Version 7 Unix and later System III. I could easily imagine that people in Commodore saw these and saw an opportunity to make them cheaper, but the Z8000 was really the wrong horse to back, especially by 1984.
@@tahrey Just to clarify. The Olivetti M20 was the computer with the Zilog 8001 CPU. The AT&T6300 is the american version of the Olivetti M24 and that has the Intel 8086-2 running at 8 MHz ... You can see both of them explained on my channel
@@TiBosRetroComputers aha, thanks. I do tend to get similar model numbers mixed up.
....maybe a bit out of nowhere, but can you scope the H and V sync lines on the video to get the frequency for each of them by any chance? Been trying to flesh out the list of such things for old machines and the documentation for them is very sketchy and unreliable. Like you do the maths on the alleged CRTC register settings and get a completely nonsense result.
@@tahrey in one of the episode is f the M20 I mentioned it. Just can’t remember now 😊
Tremendous work. Bravo! It seems a shame this design was cancelled when it was ready to go, but it is easy to forget just how crowded and diverse the field was at the time. Even a complete system represented a risk to the manufacturer.
Using a Z8000? Wow, I remember the Olivetti M20 using that chip. It was the only personal computer I knew of which used the chip.
I my self currently have 2 different Olivetti M20 computers both running. One of them even features the intel 8086 add on board extending the #Olivetti also to be able to run DOS
Absolutely fantstic! loved this video
Thanks for covering the exposed eeprom immediately for my OCD.
I used to work with Zane Healey at Intel in the 90s! He's the only person I ever knew who had multiple working SPARCbook (tadpole) laptops.
love the history videos and looking forward to many more... have a great weekend..
Very good presentation. Even as a collector, I had no idea that the 900 existed.
nice one fella, always a joy
After seeing this, there's no question Neil is keeping up with the Commodore.
Great video Neil.
I find these rare machines a real curiosity in a sea of familiarity.
Wonderful stuff! 😎👍
At uni on my A1200+Blizzard A1230+882+SCSI card, I ran Debian Linux, but that ran really hot and would just cut out when using it. It needed a bigger case, if only they'd have built tower cases for those machines and at a reasonable price, it wouldn't have been so bad.
Watching this reminded me that I have actually had a play with one of these machines at a Commodore conference that was launching the new updated business PC's (I worked for a Commodore business partner), it wasn't mentioned in the talk but was up and running with a couple of termnals.
Wow , amazing.....first time that I hear about this machine......one wonders how it could have been ....great video congrats..I hope the specialists can fix it, I'll like to see more of it 🙂
FWIW: Gerard Bucas (former Commodore Exec at the time) was very much involved with the C900 development. Very interesting machine and chips.
This is a great insight into a machine that deserved to be released to market at the time. I think it would have sold well. I used UNIX based machines in college in 88 and 89 and I found it to be very usable and extremely stable. So UNIX was a good move I think for the time. What could have been, eh? . Such a great shame. But this technibble provides a lot of great information which is excellent to learn about this machine and also to actually see an example. Thanks Neil..
14:33 I have one of those St-225 drives. Still works great :)
it is beautiful
Nice looking machine and likely very capable for the period.
Very interesting. Looking forward to seeing it run
Case looks very similar to the PC10 which we used in school for learning programming with Turbo Pascal.
Also very similar design language to the A2000, they all probably came from the Braunschweig branch.
@@skabde Yup!
A very interesting history lesson. Thanks Neil!
Nice work and super interesting machine! Too bad it got scrapped. I can confirm that the keyboard is a standard Commodore PC keyboard, it's the exact same model as the one that originally came with my Commodore PC 10-II, also with the German layout.
I would love for you to do a video on the Amiga Transputer which Commodore showed off at the NY Commodore show in the late 1980's. It looked really interesting, but it was never produced and marketed.
Are you confusing that with the Atari Transputer workstation or did Commodore have one as well.
@@MrDuncl It was originally the Commodore Transputer for the Amiga, and when Commodore dropped the project, they took it to Atari.
@@teekay_1 Well, if by "They" you mean Tim King of Metacomco, maybe. The "Commodore Transputer Project" was simply a board designed for the Amiga 2000, at our engineering office in Braunschweig, Germany, some years after Jack Tramiel and his associates had left the company. Tim was working on a new company (Perihelion Software) and operating system (HeliOS) for the Transputer and apparently evangelizing it around the industry. Atari's Abaq was a complete packaged system internally based on a modified ST plus a Transputer module, but far as I know there was no connection to the project done at Commodore.
@@DaveHaynie Interesting Dave. I'd read somewhere differently. But you were there so you know what actually transpired. I did see a demo in NY I believe sometime in the late 80's at an Amiga Expo. I thought it was pretty cool.
Fantastic. I have never heard of the 900.
Excellent video, I have never heard of the Commodore 900 before! This is one fascinating machine, I really hope you can get it up and running as curious to see how it looks and what it can do. Next question is software for it ...
It just looks so nice!
I remember seeing the 900 in the `85 Commodore magazine. It was a few years later when I owned an Amiga 2000, I kept wondering why I had felt I had seen this thing before. Then I later found my copy of the mag and had that ah-ha moment. Also… It is said that the Atari ST is the true legacy of this machine. I've read accounts that the engineers that followed Jack Tramel over to Atari, brought the 900 plans over with them. Z8000 replaced with MC68000, CPM/68k instead of Coherent.
Is it, though? The ST used a very different set of off the shelf chips, as well as the different CPU. It’s always felt to me like the STs were an attempt to produce a cost-reduced version of the Mindset PC, but with a 68000 and Mac-like OS.
(There was also talk at the time of Atari going with National Semiconductor’s NS32000 chip, but it was ultimately determined the 68000 was faster.)
Yeah, I was thinking the same all the way through the video, from fairly early on. I even had a bet that there would be a 640x400 video option and let out an audible "ah-ha!" when that was revealed... Plus there's the blitter, daisychain peripheral interface, 512k (or 128k / 2M) memory bank size, etc, though it took a while before they made a pizzabox machine, and nothing properly Unix compatible or with Workstation grade hi-rez until the TT in like 1989. I do now need to go look up the one low quality photograph I found of the vaporware "EST" from like '86-87 that would have had slightly better video and an 68020, so may have had a better chance. I have a feeling that the case looked VERY similar to this...
Plus of course Shivji was involved in both, and the ST did get whizzed up remarkably fast ... with a bit of a make-do soundchip. And a VT52 terminal emulator as a default pack in utility, and a serial based debug / diagnostic feature...
Even the first OEM hard drive was a 20MB ST225 based item, though an external one.
Though I guess the original floppy was only 360k instead of 1.2M, before going up to 720. And there was the odd throwing-in of cartridge, MIDI and mouse... and the shared video memory instead of a separate card...
Whether you'd even need full plans, because it's not the most complicated block diagram to remember, though? Just having been involved with the project would probably give you enough to go on, if you were in the position of heading up a development team at a major computer manufacturer. Maybe some of the low level stuff involved with actually making the raw low level electrical stuff work with the relatively wide and high speed bus etc might be valuable, but the general setup is fairly straightforward. More than a bit of the ST itself can probably be traced to a Motorola reference design for an S100 system, just with the video taking advantage of the higher speed memory that became available, so the C900 parts may be more conceptual than specific.
Plans for the blitter would have been valuable however... just it took a little time to get it to a suitable state for manufacture then inclusion in the Mega and STe. And maybe there's some shared DNA between parts of the C900 video systems and the ST Shifter? And then into the TT as well? (clock multiplication and specialist high speed line drivers for extra high rez video, with the advantage of an 8 rather than 6MHz basis, so you can squeeze out some extra pixels for 1280x960 instead of 1024x800...)
@@sunspot42 Maybe they gave those other chips a look-in because the Z8000 had proven lacklustre in testing and they didn't fancy repeating the same mistake? And the 68000 had already proven itself in the Lisa, which apparently was used as a dev tool for the ST, so it was kind of expedient, as well as having good reference designs and support chip families available.
The Z8000 really is a bit of a weird choice, especially for either of those two companies which hadn't done anything with Zilog (or the Intel "inspirations") before, and instead used opposite-way-up 6502s (...derived from the 6800 of course, which the 68000 retained some support / peripheral support with). The 68k really was a much more logical progression. Plus the advantage of not having segmented memory, running at the rather more optimal-for-the-time 8MHz without needing to pay for and waste a bit of something binned for 10, etc.
But, there were a lot of similar designs around at the time, if you were trying to make something affordable with meaningful 16-bit power at the time, your choices for processor and system design were kind of limited. Even those which used alternative CPUs (like the Archimedes) ended up bearing an uncanny resemblance. The comparisons are inevitable. It's just that given the personnel and other items of heritage, plus certain features of the design, made me sit up and go "hang on a minute" even not very far into the video. The ST sort of seemed like a bit of a true C64 / C128 successor if anything (vs the Amiga being seemingly more related to the Atari 8-bits), but lacking a little on the audio and hardware video acceleration (and lack of full ANSI 80-column colour) front. With the C900, a few more pieces fall into place. There's bits of each in there, conceptually.
It reminds me of the Amiga 2000 and Amiga 1500 in appearance. Good luck with getting it to work!
I used to own a Z8001 Zilog Unix (ZEUS) machine, great looking toy, but even at the time was somewhat limited. The speed was not great and the limited RAM made it pretty useless for anything other than small compiles and text editing. I think the cost of enough resources to make Unix usable was the real reason why a lot of machines like this got scrapped in development, it would be quite some time before the capacity and cost of RAM made (even a limited 16 bit) Unix worthwhile.
When I first saw the thumbnail I thought I was looking at an amiga 2000. Very similar design.
Compute! was a US based publishing company. They published computer books and magazines. ABC Publishing (America Broadcasting Company) was their parent company.
"Or in german, ein buffet mit RS232" 🤣
Depending on price and peripherals, this could have replaced mini computers, the mini I was using was configured in a similar way, but with much less memory & only 5m of hard drive space. So I could see this as upgrading small businesses at the time. Could have been interesting.
I had no idea about the 900 or the UNIX variants of the 2500 and 3000. I had both the 2500 and the 3000 which did have the 68XXX memory management chips, but Commodore never did anything with segmenting up memory. Instead, all of the apps could write all over each other.
The MMU in the A2500/20 and A2620 was entirely there to support UNIX. The later machines got it for free with the 68030 and 68040 processors. The time wasn't right for a protected AmigaOS, based on the need to support lower-end systems, which were the majority of the Amiga computers Commodore sold. However, the software team and technical support group released various MMU-based development tools (Enforcer, MemMunge, etc) that worked to help find bugs that could cause catastrophic failures. This was also encouraged enough that code reviewers caught on and generally ran these tests for reviews... you didn't want a reviewer to report Enforcer hits with your new software!
Super-neat. I love obscure and never-quite machines like this. Given that the Z8001 never really went anywhere, I wonder if the lack of adoption generally was part of the decision - Zilog might've ended up having a chip widely used by only one maker. That's in a bad place to be, if you're talking about moving forward with an architecture, something which probably wasn't real obvious in 1982 or 1983, but was starting to become quite clear in 1985.
About the Italian translation at 17:09. The word polloci it's wrong, the right one is pollici (as also suggested by the translator program). I presume a typing error during the input for translation.
The typing is correct, the original translation is wrong, you'll see it on the 900's screen later in the video when we see it being demonstrated
Thanks, I'd never heard of this machine before and it's really interesting! BTW what happened to part 2 of the Enterprise computer video? Part 1 was over a year ago now.
The connector for the blue/red/yellow cables seems burned @19:50
Would be great to see further information emerge as a result of the attention this video will undoubtedly garner.
Compute! Magazine was headquartered in Greensboro, NC (the "Canada" reference on the magazine cover is the Canadian dollar price). The domestic magazine market for the USA actually includes Canada and the US, as does the domestic box-office/movie market.
Thanks! Someone else also pointed this out I’ll mention it in the next video 👍
@@RMCRetro My pleasure. I am learning so much about PAL and UK computing / video gaming through your channel!
What a beautiful machine - and I hear Dave is a really lovely guy :D
There are quite some informations about this machine in Brian Bagnall‘s Commodore book („Commodore - A Company on the Edge“, in Germany the book is sold under the name „Volkscomputer“. See page 258ff in the german edition about the evolution and canceling of the machine.
Thanks I'll find a copy and share any further info I find in it when/if we get to testing this out
I remember the VIC-20 (aka VC-20) being marketed as the "Volkscomputer" in Germany, The Netherlands and Belgium.
I also remember waiting for home computers with the Z8000 and Z80000 "mainframe on a chip" CPU's to become available during the early to mid-80's and killing the inferior 8088 IBM PC (which I considered to be "too little and too late" - boy, was I wrong!), but this never happened.
@@koenlefever The "VC" came of course since "VIC" spelled out in German sounds just like the word for "fuck". So they changed it to VC. Ret-conning it into an abbreviation for "Volkscomputer" was probably just an idea by the marketing guys...
It would be interesting to see if that keyboard is pin-compatible with the Amstrad PC 1512/1640. The layout looks exactly the same and it has the same Lamps-under-keys scheme for Caps/Numlock
What a quaint little machine. Even comes with mysteries and a brief course in German!
The 68k architecture is 32-bit, as the instructions and registers were 32-bit. The 68000's implementation relied on hardware which could only operate on 16 bits at a time, so it came with a performance penalty, but the fact remains that as far as software was concerned, it was no different than a 32-bit clean CPU.
...except for the 24-bit memory addressing, you mean? ;)
@@tahrey yeah but would ever have a need for so much RAM that one would have to use all those 32-bits :-)
@@Henrik_Holst The 24-bit address allowed for some bad programming habits. For example, Microsoft BASIC was using those unneeded top 8 address bits for some kind of extra data fields, so it failed hard once we had machines out with memory beyond that first 24-bits which, of course, required 32-bit clean code. Once we had machines with MMUs, developers could run Commodore's Enforcer system and test all this even without having that extra memory.
@@DaveHaynie of course people would abuse that :). The same happens today with 64-bit addressing since no one imagines that we will ever see such a large system. If I recall correctly e.g Intel uses up to 15-bits for pointer tagging.
Everything repeats :), and oh btw huge thanks for your work on the Amiga that and the c64 was what teached me programming back in the day (more or less was forced to on the c64 since I didn't had either the datasette or a floppy initially).
Was the CPU in that machine compatible with the ZILOG Z80A processor.
no, at least not binary-compatible. the Z8000 supported equivalent instructions (it could do the same things) with equivalent memory addressing (it could get/put things the same way) but the two were not directly compatible.
working from memory, a source-code translator was available that would map the Z80 instructions to Z8000 instructions, which could be assembled and linked to create an equivalent (but different) executable program that behaved the same way as the origina Z80 program.
@@standroid2406 Sorry.
@@standroid2406 OK No problem.
It's amazing how functional systems were back then with so little RAM.
It shows how little room is needed for text and numbers. When pictures and videos cropped up, then we needed way more RAM (and larger removable disks too e.g. SyQuest, IoMega). As nice as the C900 might have been, my friend and I just ran Xenix on Compaq Deskpro 386s and added multiport serial cards. Maybe Commodore realized that it wasn't worth competing against such commodity "micro mainframes".
Or it shows how inefficient and bloated modern OS and software has become.
Those of us programmers that learned to code on a Vic-20 with 3.5KB of RAM learned to optimize code/data/etc. Sadly, this is mostly a lost art now and hence the bloat of most software.
@@GeoffSeeley very true but you trade that for APIs that allow software to work across a huge variation in hardware. Most modern software could probably run on a Core 2 Duo 4GB system in a Windows 7 PC and on an AMD 64 Core Threadripper 64GB system in a Windows 11 PC. Back then you had to tailor your software very specifically for each hardware.
Edit
I'd say the changes are better for developers to actually make money, as software gets bigger and bigger, being able to sell to more people gets more important.
No, the lack of memory always made the systems primitive. The text-only interfaces, hardware monospace fonts, single-byte character sets, US-centered software...
Byte Magazine! ❤
Had to lough when you read "tot" - Reminded me a little bit of a Simpson episode "Die Bart Die"
Lovely clean lines, just like my Commodore PC5
What's the name of the music played in the beginning of the video. Very ambient chill.
Music in this video is:
1AM OMW - Ballpoint
All Parts Equal - Airae
High Octane - Hampus Naeselius
Innovations - From Now On
Let There Be Rain - Silver Maple
Lofty Palms - Yi Nantiro
Message Between Worlds - William Benckert
Revitalize - Yonder Dale
The Farmhouse - Silver Maple
@@RMCRetro Silver Maple - Let There Be Rain. Beautiful tune, Thank you.
@@treychatillon6994 Makes me appreciative how much effort is put into these videos when he uses 9 different tunes for a 30min video on a niche subject. He had to sample multiple music tracks to find the ones he wanted to use, or at least have already sampled them to have in his library of acceptable music to use.
I had a Z8001 machine back in the early 80's from Whitechapel Workstations. This was a British competitor to SUN 2 workstations and similar Unix workstations of the time.
I visited the company base in Whitechapel London for demonstrations a few times. The machine had a high definition black and white monitor for CAD and similar applications. They worked quite well, but it was very difficult for the company to compete against Sun Micro Systems and the Motorola 68000 ubiquity in the CAD market. I seem to remember some hardware/firmware limitations/bugs in the Z8001 chips in the virtual memory area. This was not helped either by Zilog as a company not being stable. I don't think the Zilog 8000 processors ever got much of a market. Perhaps Commodore hit the same issues.
Whitechapel's MG-1 was a 32016-based workstation, and the NatSemi chipset had quite a few problems that inhibited adoption of the 32000 series. Indeed, there are signs in reviews of the MG-1 that WCW might have had to use some workarounds for problems with the memory management unit. Acorn Computers were supposed to deliver Unix on the 32016 but never did, with the MMU also featuring as an obstacle in their plans, although one can argue that Acorn didn't seriously pursue that product and didn't commit the resources to it, either, going through near-bankruptcy at that time.
@@paul_boddie Thank you Paul for the correction. I had completely mixed up the two processors and companies.
@@Ray-ds5dc No worries! Great to hear that you used an MG-1 back in the day, though. I hear it was briefly popular in London financial institutions.
As for Z8000-based machines, reading up on them a bit, I think that one of the first ones to reach the UK was the Onyx C8002. That was actually based on the Z8002 with custom memory management hardware because Zilog's MMU appears to have been delayed, this pushing back the introduction of Z8001 machines. The C8002 seems to have been competing with PDP-11 machines, which would not have been a viable strategy for very long, however.
So, there are definitely parallels between Zilog and NatSemi in terms of problems delivering their chipsets. Interestingly, vendors shipping 68000-based systems also developed their own memory management hardware until the 68010 came out. At that point, the Z8000 would not have been very attractive, and by 1984, when the 68020 came out, it seems that various companion chips for the Z8000 were still not available. There would have been many reasons at that stage for anyone using the Z8000 to jump ship to something else.
... Remember ... once upon a time there was a multi user multitasking operating system
written in BCPL called TRIPOS ,, from Cambridge University ... it was ported to the 68000
and used in a machine called the ... Amiga
.... so,, a business system using the early version of Amiga OS & serial terminals, was already in operation ...
I wonder if the front provided inspiration for the front of the A2000?
Great video, thanks Neil.
Looking forward to hearing if it can be resurrected.