Some thoughts on why the Amiga didn't thrive

[Signman]

[saimon69]

[Foebane]

Originally Posted by Signman

Originally Posted by saimon69

What are you trying to say? That we Amiga zealots are our own worst enemy?

[saimon69]

Originally Posted by Foebane

What are you trying to say? That we Amiga zealots are our own worst enemy?

Worst and fiercest sometimes...

[Frogs]

Originally Posted by dreadnought

So, as I was saying on page 2...

. But given a few different decisions early on, we might be using Amigas instead of the Macs. Apple wasn't exactly the best run company in the late 80s, early 90s either.

[Bruce Abbott]

Originally Posted by Keatah

There is absolutely no shame in wanting compatibility and interoperability. Wanting different just to want different (and make things harder) is ridiculous.

You were thinking of Windows 10 when you wrote that, right?

Not interested in going through gymnastics to get media working, or viewing photos that were given to me on a USB stick that I want to archive.

Yeah, all those different video and picture formats that people created just to be different (they will tell you it was to make it better, but really it was just so they could say "hey look at me, I made something different!"). Luckily the Amiga has Datatypes, so anyone can write a viewer for whatever format is created (provided the specs are public).

File format standardization is good for everyone. Remember the bad old days when people would send you 'text' documents that were actually in some proprietary format created by a word processor you didn't have? Or videos encoded with DivX or Window Media player that nothing else would play? And the clueless users who tried to send you raw images straight out of their camera in max resolution, crashing your email server?

One of the first personal computer companies to recognize the importance of standardized file formats was... Commodore! In 1985 - in conjunction with Electronic Arts - they created the Interchange File Format (IFF), which was later adopted by IBM and Microsoft for their Resource Interchange File Format (RIFF) used in AVI and WAV files, and also as the container for Google's WebP picture format.

Not interested in wrestling with re-formatting a document when it comes time to print.

I know what you mean. My Dell PC came with OpenOffice, which also came with the version of Linux I use - so life was good. Until my brother's girlfriend asked me to print out her CV (because they were too stingy to buy their own printer). Being clueless she carefully adjusted the document in Word to get the text where she wanted it, using extra spaces and newlines etc. Of course when I loaded it into OpenOffice the formatting came out different, and it took me ages to get it back to how she wanted it.

Now some people would say that's my fault for not having the same program she did, or having A4 instead of letter size paper etc., but why should I have to spend up large just so she doesn't have to learn how to use a word processor?

Then there are the spreadsheets people create with fancy fonts that I don't have on my PC, so OpenOffice tries to substitute the best it can and it comes out a mess. If only they would stick to the stock fonts or send me a csv file it wouldn't be a problem! But that's the thing - people want different just to be different. Stock fonts are boring, let's use this fancy one that other people probably don't have!

[Keatah]

Originally Posted by Bruce Abbott

But Commodore had never operated like that. They always just made hardware and expected it to sell itself. PC clone manufacturers did the same of course, but they were part of an ecosystem that provided the software and support customers needed. Riding on the coattails of IBM and Microsoft made it so much easier for them.

When I got my Amiga BITD it was because I wanted to upgrade to 16-bit computing. Something more capable than the Apple II.

I had considered the MAC, way too expensive. IIgs also too expensive, but a buddy of mine got one. Considered the ST, but I don't believe it was available in my area for cheap. Only thing I could afford in '87 was the Amiga, and that included a special sale price. At no point did I consider a PC. It was automatically too expensive with all the necessary upgrades. And it was complex looking. And there were no friendly tutorials or vibrant color manuals. It was too much for me.

To my then-infantile mind the specs of all machines were similar. And since I was interested in 16-bit computing for graphics/art, any of them would do. Amiga won out on price. And I became thoroughly enamored with DeluxePaint, PhotonPaint, and DigiPaint. Had all three in my arsenal along with Digi-View digitizer. And away I went drawing up something every couple of weeks. Moody raining nights. Incredible fun!

The Amiga and Apple II co-existed side-by-side till around 1992-1993. It was then that I was outgrowing the stock Amiga 500's color palette and resolutions. I also had tried to transition my personal and professional writing over to the Amiga. But that didn't go over too well.

In those early 1990's I got wind of all the cool astronomy and scientific programs for PC. Games were an afterthought. I immediately wrote in for catalogs and settled on a Gateway 486 for around $2000 - $2500 depending on exactly when I got some additional peripherals.

It ran at a workstation-class speed of 50MHz and came with a monster-sized 200MB HDD. This was Big Boy stuff! And I immediately discovered I could trade files with just about every business in town, as well as the millions of other PC owners. It was liberating.

I got the PC mainly because of the available software. I was too young yet to appreciate the bandwagon of upgrades that was possible. But I did understand the 8 16-bit ISA slots. The more slots the better.

Originally Posted by Bruce Abbott

But don't kid yourself. The PC juggernaut was crushing everything, and nothing could stave off the inevitable for long. When 486 clones hit the UK it was obvious where things were headed, and then Microsoft released Windows 95...

I was never totally impressed with Windows95, just having learned 3.1. But I upgraded and not long after made the jump into Pentium II. This time it was all about games and graphics.

[Foebane]

As I only really used my real Amiga for demos, I wonder how things would've panned out had Doom not caught my attention.

I had an A1200 that was unexpanded except for an 85Mb HDD, and it was great for that purpose, although back in 1993-4 I would've soon be limited in my upgrade options for demos.

FastRam, for instance - some A1200 demos needed it, and so that would be a problem, not to mention potential upgrades that involve opening up the case and poking around inside, something I was squeamish about and had never done before.

As it turns out, PC ownership was very expensive for me too: After having gotten a 486 SX2/50 with multimedia gubbins, I very soon had to expand the 4Mb I had to 8Mb, and I managed to fry my first £100 such expansion by improperly handling it (I think I got the replacement for free). Then I had to upgrade the CPU to a DX2/66 so I could play Quake, or attempt to, and I think the shop did it for me.

Then there was the time I took out a student loan for my first 3D card upgrade! I've been deferring repayments ever since, because I've not got a huge wage, and luckily (hopefully) it expires next year, as it's been 25 years and I took it out in 1997.

Not to mention that on average, I've bought a new PC every FOUR years! Now that is expensive!

[malko]

Originally Posted by Bruce Abbott

[...] Then there are the spreadsheets people create with fancy fonts that I don't have on my PC, so OpenOffice tries to substitute the best it can and it comes out a mess. If only they would stick to the stock fonts or send me a csv file it wouldn't be a problem! But that's the thing - people want different just to be different. Stock fonts are boring, let's use this fancy one that other people probably don't have!

[small OT]I have substituted the Wind@ws default font with this one : DejaVu. People should give it a try if never heard about it.[/small OT]

[grond]

Originally Posted by Bruce Abbott

Producing chips with good yield becomes increasingly difficult as the transistor size reduces.

This is not correct. Yes, it becomes increasingly difficult to produce chips as the transistor size reduces. However, yield actually improves with shrinking because the same design on a smaller area is less likely to hit a defect in the silicon wafer. In fact yield is inversely exponentially proportional to the chip area of a single chip (Y~e^-A*D with Y yield, A area and D defect density).

Each process generally requires a dedicated production line, which costs millions of dollars to set up.

And here we go again: while a smaller node size improves yield (see above), saves expensive wafer space (more chips fit on a wafer) and thus makes manufacturing cheaper, Commodore didn't want to invest money into the fabs because they saw investments as a waste of money. They preferred short-term profits over long-term profitability. If I'm not very much mistaken, MOS actually once had quite an attractive product portfolio including one of the most popular processors of its time but managed by Commodore stagnated and eventually died just like every thing Commodore.

So in 1991 (when the AGA chipset was designed) they probably could have done better than 5 µm. But that would have required a complete redesign of the custom chips to use the new process

Not necessarily. Usually the first thing to do when you have updated your process node is to produce the same layout as before (not shrunk). Yield already becomes better because the new process manufacturing the old layout is more precise and thus manufacturing variations are less likely to render a chip faulty.



The next thing is to do a shrink (basically geometrically scaled) and only then you redesign. However, there was another problem with Commodore's stoneage (because never updated) process: they couldn't produce CMOS chips, as they lacked the extra diffusion layer. A redesign of the custom chips to CMOS was desperately needed to reduce power consumption and, IIRC, was eventually done for some of the AGA chipset that was manufactured by HP (Lisa?).

[saimon69]

Originally Posted by Bruce Abbott

I know what you mean. My Dell PC came with OpenOffice, which also came with the version of Linux I use - so life was good. Until my brother's girlfriend asked me to print out her CV (because they were too stingy to buy their own printer). Being clueless she carefully adjusted the document in Word to get the text where she wanted it, using extra spaces and newlines etc. Of course when I loaded it into OpenOffice the formatting came out different, and it took me ages to get it back to how she wanted it.

Now some people would say that's my fault for not having the same program she did, or having A4 instead of letter size paper etc., but why should I have to spend up large just so she doesn't have to learn how to use a word processor?

Then there are the spreadsheets people create with fancy fonts that I don't have on my PC, so OpenOffice tries to substitute the best it can and it comes out a mess. If only they would stick to the stock fonts or send me a csv file it wouldn't be a problem! But that's the thing - people want different just to be different. Stock fonts are boring, let's use this fancy one that other people probably don't have!

Lately things are much better and i more or less can transport discretely complex documents from office to libreoffice (more updated); about not including fonts remember that some typefaces want $$$ to be licensed so they provide you those similar ones in case you have no fonts at all in the machine (VERY rare lately but still possible)

[TEG]

@grond It's very interesting.

According to Wikipedia, MOS was bought by Commodore in 1976 and was a Commodore property until 1994: 18 years!

Do you have an idea when the "extra diffusion layer" technologies appeared? And does it refer to this?:

Some dopants are added as the (usually silicon) boule is grown, giving each wafer an almost uniform initial doping. To define circuit elements, selected areas — typically controlled by photolithography — are further doped by such processes as diffusion and ion implantation, the latter method being more popular in large production runs because of increased controllability.

Small numbers of dopant atoms can change the ability of a semiconductor to conduct electricity. When on the order of one dopant atom is added per 100 million atoms, the doping is said to be low or light. When many more dopant atoms are added, on the order of one per ten thousand atoms, the doping is referred to as high or heavy. This is often shown as n+ for n-type doping or p+ for p-type doping. (See the article on semiconductors for a more detailed description of the doping mechanism.)

source

There is an interesting part in the MOS Wikipedia page:

MOS had previously designed a simple computer kit called the KIM-1, primarily to "show off" the 6502 chip. At Commodore, Peddle convinced the owner, Jack Tramiel, that calculators were a dead end, and that home computers would soon be huge.

However, the original design group appeared to be even less interested in working for Jack Tramiel than it had for Motorola, and the team quickly started breaking up. One result was that the newly completed 6522 (VIA) chip was left undocumented for years.

Bill Mensch left MOS even before the Commodore takeover, and moved home to Arizona. After a short stint consulting for a local company called ICE, he set up the Western Design Center (WDC) in 1978. As a licensee of the 6502 line, their first products were bug-fixed, power-efficient CMOS versions of the 6502 (the 65C02, both as a separate chip and embedded inside a microcontroller called the 65C150). But then they expanded the line greatly with the introduction of the 65816, a fairly straightforward 16-bit upgrade of the original 65C02 that could also run in 8-bit mode for compatibility. Since then WDC moved much of the original MOS catalog to CMOS, and the 6502 continued to be a popular CPU for the embedded systems market, like medical equipment and car dashboard controllers.

The reputation of J. Tramiel seems to be not good in business. If I remember well, Atari guys were unenthusiastic when they learned who will buy the company. I have to found back the video.

In CACOTE Peddle explained he had setup the first CMOS sensor but Tramiel shut down the research and the R&D lab because it was not something he had authorized to research on. This is on the line of what you say.

Finally with the distance, it's appear that Commodore, on the long run, paid the price of it's behaviour which made it a millions dollars company for a time.

[grond]

@TEG: a second diffusion layer is what you need to progress from NMOS to CMOS technology. CMOS is a thing of the late 1970s and has the big advantage of being vastly more power efficient than NMOS (heat problems in Denise anyone?). Your above info on Bill Memsch shows that the step to CMOS was the logical next step and a small design group moved the MOS6502 to CMOS. Something that Commodore never did but could have done. Commodore might be today's Intel if they had managed MOS better.

[Bruce Abbott]

Originally Posted by grond

This is not correct. Yes, it becomes increasingly difficult to produce chips as the transistor size reduces. However, yield actually improves with shrinking because the same design on a smaller area is less likely to hit a defect in the silicon wafer. In fact yield is inversely exponentially proportional to the chip area of a single chip (Y~e^-A*D with Y yield, A area and D defect density).

If this was true then chip manufacturers would have gone for the smallest possible size straight away - but they couldn't because their processes were not clean and accurate enough. Only after spending a lot of money and effort developing a better process to create smaller structures could they look at shrinking the die size or putting more stuff on it. Trying to do it with a cheaper and 'dirtier' process would result in much worse yields.

Semiconductor fabrication plant

Fabs require many expensive devices to function. Estimates put the cost of building a new fab over one billion U.S. dollars with values as high as $3–4 billion not being uncommon. TSMC invested $9.3 billion in its Fab15 300 mm wafer manufacturing facility in Taiwan. The same company estimations suggest that their future fab might cost $20 billion.

The central part of a fab is the clean room, an area where the environment is controlled to eliminate all dust, since even a single speck can ruin a microcircuit, which has nanoscale features much smaller than dust. The clean room must also be damped against vibration, to enable nanometer-scale alignment of machines, and must be kept within narrow bands of temperature and humidity...

Another side effect of the cost has been the challenge to make use of older fabs. For many companies these older fabs are useful for producing designs for unique markets, such as embedded processors, flash memory, and microcontrollers... older foundries can be cheaper to operate, have higher yields for simple chips and still be productive.

Originally Posted by grond

And here we go again: while a smaller node size improves yield (see above), saves expensive wafer space (more chips fit on a wafer) and thus makes manufacturing cheaper, Commodore didn't want to invest money into the fabs because they saw investments as a waste of money.

All business is a balance between investment and profit. Commodore was not in a position to pour huge amounts of money into developing denser chips. Intel could, and in fact had to in order to keep ahead of the competition. But Intel had developed a market where they could convince customers to pay ridiculous prices for the latest CPUs.

That's not to say that Commodore didn't invest some money into 'the fabs'. Between 1985 and 1994 they shrunk their process from 5um to 1um - not as much as the industry leaders, but not 'stagnating' either. And just because you have a smaller process doesn't mean you can switch your existing designs to it. Smaller transistors have different characteristics that often demand a complete redesign of the chip. Many chips today are still produced with older processes because it's not the worth the effort (and probable incompatibilities) to make them smaller.

So Commodore had a choice of redesigning the Amiga chipset to use the smallest process they could, or sticking with something that they knew worked. You talk about 'stagnating', but the design process is not instantaneous, and in those days was a lot slower than it is now. By the time you had developed and debugged a wire-wrap prototype using standard logic chips and/or emulators, transferred it to silicon, and manufactured enough chips and machines to test and debug them (possibly going through several iterations) several years could have passed - years that you don't have because you need machines out there now to develop a software base.

Commodore already had experience with buggy chips that crippled their designs. The 6522 had several bugs, one of which prevented use of the shift register for high speed disk transfers. This bug was fixed in the 6526 used in the C64, but to cut development time and maintain compatibility they continued to use the slower bit-banged interface. If Commodore had held the C64 back until they got all the bugs out it probably wouldn't have dominated the market, and the Amiga might never have happened.

And they were not the only ones. Acorn had problems with the custom chip in their Electron that caused them to miss the Christmas period, which was a critical marketing failure.

Reports during the second half of 1982 indicated a potential December release, with Curry providing qualified confirmation of such plans, together with an accurate depiction of the machine's form and capabilities, noting that the "massive ULA" would be the "dominant factor" in any pre-Christmas release. However, as the end of the year approached, with the ULA not ready for "main production", the launch of the Electron was to be delayed until the spring...

By June 1983, with the planned March release having passed, the launch of the Electron had been rescheduled for the Acorn User Exhibition in August 1983...

As the company increased production during 1984, however, the British home computer market greatly weakened.

Imagine if Commodore had been caught in a fiasco like that! They deserve praise for producing solid reliable machines with no major bugs, and not trying to 'push the envelope' too far. Of course PC clone manufacturers did the same only more so - generally only introducing incremental changes to make a profit now, rather than pouring funds into future 'game-changing' products that might miss the market.

They preferred short-term profits over long-term profitability.

A company exists for the sole purpose of making money for its investors. Who was the biggest investor in Commodore? Irving Gould.

In 1965, Jack Tramiel, Commodore's founder and CEO, decided to purchase the Canadian store chain Wilson's Stationers to provide a sales channel for their products. To fund the purchase they borrowed $3 million from Atlantic Acceptance Corporation at an 11% interest rate. On 14 June 1965, Atlantic bounced a $5 million check and was insolvent within days. This led to all their capital loans being called in, including Commodore's $3 million.

Looking for a way out of the problem, Irving Gould arranged the sale of Wilson's Stationers to a US company. To pay off the bridge loan, Gould purchased 17% of Commodore's stock in 1966 for $400,000. Over the next decade, the company repeatedly had difficulties and repeatedly turned to Gould for funding...

The Japanese companies were able to undercut Commodore both in technology and by being vertically integrated. Texas Instruments, one of Commodore's suppliers, decided to follow this pattern and introduced complete calculators at prices below what they sold the parts to Commodore. Gould provided funding to keep Commodore going during the period where they were being forced out of the calculator business. Tramiel responded by buying MOS Technology to supply microprocessors and moving into the computer market.

As you know, Tramiel later gutted Commodore when the board wouldn't let him have his way, leaving Gould - who had been pouring money into it all this time - with a shell of a company. So naturally 'short-term' profits were a priority. To much 'long term' efforts at that time would have just killed the company off sooner, leaving Gould and the rest with nothing - and us with no Amigas.

By 1994 Irving Gould was 85 years old and looking to cash out - which as fair enough - and not a moment to soon because the market Commodore had done so well in (home computers) was almost dead. Perhaps if Gould had been a more savvy businessman he might have employed smarter people who could have extended Commodore's profitability for a few more years. But there's a good chance that path would not have included better Amigas.

If I'm not very much mistaken, MOS actually once had quite an attractive product portfolio including one of the most popular processors of its time but managed by Commodore stagnated and eventually died just like every thing Commodore.

Commodore bought MOS to make chips for their computers, not to supply the general market. This worked well for a time, allowing Commodore to undercut other manufacturers on price and performance - and is arguably the main reason the C64 became the best selling single computer model of all time.

But that was the early days when a only few manufacturers were in the game, each producing their own architecture. By the 90's everybody was making PC clone stuff and the race was on to do it better and cheaper. In that environment doing 'long term' development on an alternative platform was bound to make you fall behind. Even the big boys failed when they tried it. IBM poured billions into PS/2, and Intel did the same with iAPX432 - to no avail. You either churned out stuff for clones to make profits now, or perished.

However, there was another problem with Commodore's stoneage (because never updated) process: they couldn't produce CMOS chips, as they lacked the extra diffusion layer. A redesign of the custom chips to CMOS was desperately needed to reduce power consumption and, IIRC, was eventually done for some of the AGA chipset that was manufactured by HP (Lisa?).

Once again, Commodore's only goal in owning a fab facility was to make cheap chips for their home computers. Their only market was themselves, so they didn't have the sales volume to justify developing new processes.

Had Commodore continued producing new Amigas they probably would have moved to mostly fab-less designs like most other chip 'manufacturers' were doing. ISTR an engineer at Commodore saying that they could go from schematic to silicon in a few weeks, rather than the months or years it used to take. By the mid 90's FPGAs with over 20,000 logic gates were also available, which could be used for rapid prototyping or finished products (if you didn't mind the price).

[EugeneNine]

I've posted before in a similar thread that I don't think Commodore mismanaged the end. I think they new the PC had already won the 'war' and made a business decision to milk the cash cow for whatever profit they could for as long as they cold before ending it. Which in reality is good business.

Big businesses had IBM midrange and mainframe in the back room already so it would have been a difficult sell to put another brand PC on the desktop. Smaller businesses had s100 systems typically and while there were varying brands of CPU's and OS's for them, the 8088 XT and MS DOS were fairly close to the 8080/z80 and CP/M that some software companies could migrate. Remember back then and even still today in the large system world software is leased/licenses, its not bought (Bill Gates gets way too much credit for simply doing things they way they were already done (licensing instead of outright selling, it wasn't a brilliant/smart thing he did, it was the safe/status quo thing he did). MS-DOS uses many of the same software interrupts as CP/M for many of its system calls and 8088 instructions are an extension of the 8080 instruction set. Z80 instruction set was 8080 with some additions. So software vendors who wrote for/used CP/M on 8080/z80's could bring to marker DOS/8088 software faster than others. 6502 didn't yet have any 16 bit upgrades, and Motorola broke compatibility between the 6800 and 68000. So if you had a 6800 based S100 minicomputer in your back office your software vendor didn't have any similar pc to port the software too. Plus many of the smaller software vendors did sell the software so they didn't have the residual income that the vendors who leased did. So those small vendors would need someone to financially back them porting their software to a new 'pc' or do it in their spare time or just not move forward. The software vendors who were leasing or were subcontracted by hardware vendors like IBM had the regular income already and business planners looking ahead for the enxt system they could lease software on so they were ready to go at te launch of the IBM PC.
Really by that time Apple, Atari, Commodore were already niche markets, big business added IBM PC's on to their existing IBM leasing agreements and kept moving. Small businesses kept nursing their old systems along of bought off lease IBM. So as the other smaller vendors started going away you ended up with IBM. The other companies saw that, Compaq, HP, Tandy, etc started with their PC compatible so new names came on the scene as the old non-pc compatible branded faded away. Meanwhile the Atari 2600's, Colecovision, intellivsiion,etc sales were dying partly because the home pc's like the Commodore 64 were eating into their sales and the software marker was flooded by poor quality releases from anyone and everyone that could write a game overnight trying to make a quick buck. The little company halled Hi-torro saw console sales lowering and pc sales increasing and quickly added a keyboard to make their console a home pc.

The tech media was already asking 'whats next from Commodore/Atari/Apple' and Commodore was already trying to sell some PC clones but I think their management knew then they couldn't compete with IBM, Compaq, HP, etc, etc. So they grabbed hold of Amiga Inc to raise their stock prices and therefore profits for a bit and then kept Commodore on minimal life support and just kept the profits until the end. Thats how business works, the goal is profit and profit is more about the stock price than the unit profit on each widget you sell.

[Charles Littell]

I fell that a lot of people are missing something, before Commodore went out of business the economy was in tough shape, other companies went of of business, Atari for one, many PC compablitable companies. Apple even almost went out of business, it was bad. I remember around 1990 or 1991, Commodore updated their operating system and then laid many of those people off. I lived out in Hawaii at the time and we had two stores selling Amigas and they were flying off the shelf. I really think that Commodore just was trying to do everything on the cheap and in the end it bite them in the ass. One last question why the hell were they making a PC clone?

[Bruce Abbott]

Originally Posted by Charles Littell

One last question why the hell were they making a PC clone?

Same reason as all the others - to make money!

Olivetti, Ferranti, Tandy, Amstrad, Toshiba, Apricot, Mitsubishi, Canon, Atari, Epsom, Research Machines, and many more if you include machines that were not quite IBM compatible. A list of established computer manufacturers who didn't jump on the IBM bandwagon might be shorter!

[defor]

Originally Posted by Charles Littell

One last question why the hell were they making a PC clone?

Why every mobile and tablet manufacturer is currently making a clone of ARM+Android blueprint?
Software, price, demand.

[grond]

Originally Posted by Bruce Abbott

If this was true then chip manufacturers would have gone for the smallest possible size straight away - but they couldn't because their processes were not clean and accurate enough. Only after spending a lot of money and effort developing a better process to create smaller structures could they look at shrinking the die size or putting more stuff on it. Trying to do it with a cheaper and 'dirtier' process would result in much worse yields.

That's obvious, isn't it? It's not like each semiconductor manufacturer has to develop their own process, there are companies that do that. It was only towards the end of the millenium that companies had to invest billions and thus basically formed two groups - Intel vs. the others - to develop new process nodes. And even then the optics for nano-meter scale feature sizes were developed by specialist companies that delivered to both sides.

Commodore would have had to buy what was there and proven to work. Like always they considered such investments as costs that cut the short-term profits. Just have a look into an A1200 and note the size of the 020 processor and the size of all custom chips combined. It's not like the 020 is the least complex chip in the lot.

All business is a balance between investment and profit.

Yes, and Commodore always went for the short-term profits that strangled the long-term profitability.

That's not to say that Commodore didn't invest some money into 'the fabs'. Between 1985 and 1994 they shrunk their process from 5um to 1um - not as much as the industry leaders, but not 'stagnating' either.

But most importantly they never made the step from NMOS to CMOS. If Commodore had run MOS as a CPU manufacturer that was selling a very popular processor also to competitors, they would have had to make that step (see above, some former MOS devs adapted the 6510 to CMOS apparently on a comparatively low budget) and also upgrade its architecture to 16bit. If they had done that in time, who knows, their product could have become the processor of choice for the IBM PC. Similarly ARM originally was a computer manufacturer, now only the CPU survives.

And just because you have a smaller process doesn't mean you can switch your existing designs to it. Smaller transistors have different characteristics that often demand a complete redesign of the chip. Many chips today are still produced with older processes because it's not the worth the effort (and probable incompatibilities) to make them smaller.

That's not so much because using the older fab you save effort for the shrink, it is more because the old fab is already paid for and the new fab is running with more demanding chips at 100% capacity. A shrink is not very demanding. I explained the approach above.

By the time you had developed and debugged a wire-wrap prototype using standard logic chips and/or emulators, transferred it to silicon, and manufactured enough chips and machines to test and debug them (possibly going through several iterations) several years could have passed

Nonsense. You had a working design, you don't start again from discrete gates. When you use a new process, you port your library of components (NAND-gates, flip-flops etc.) over to the new process one by one. When each component works, you port the complex design over to the new library of components which is pretty straight forward. You do a testchip or two and are done with it.

Of course PC clone manufacturers did the same only more so - generally only introducing incremental changes to make a profit now, rather than pouring funds into future 'game-changing' products that might miss the market.

The made incremental upgrades when Commodore did none.

Once again, Commodore's only goal in owning a fab facility was to make cheap chips for their home computers. Their only market was themselves, so they didn't have the sales volume to justify developing new processes.

The 6502 was used in popular computers such as the Apple II. The market for it and any possible updates was surely bigger than only Commodore computers. How did markets of the MOS processors and Intel processore compare in size before the IBM compatible?

[Bruce Abbott]

Originally Posted by grond

The 6502 was used in popular computers such as the Apple II. The market for it and any possible updates was surely bigger than only Commodore computers. How did markets of the MOS processors and Intel processore compare in size before the IBM compatible?

Before 1981? Data is hard to come by, but if you include the Z80 then I would guess the markets were comparable. But it's not relevant anyway. Commodore never intended to be a chip vendor, and other manufacturers were already making the 6502 when they bought MOS technology. Why would they supply their competitors when they needed all they could make for their own machines?

But most importantly they never made the step from NMOS to CMOS.

Even worse, they didn't change most of the other chips in the Amiga to CMOS either - I wonder why? Could it be that it wasn't so important?

If Commodore had run MOS as a CPU manufacturer that was selling a very popular processor also to competitors, they would have had to make that step (see above, some former MOS devs adapted the 6510 to CMOS apparently on a comparatively low budget) and also upgrade its architecture to 16bit. If they had done that in time, who knows, their product could have become the processor of choice for the IBM PC.

When Commodore bought MOS its best engineers had already left and it was headed for bankruptcy. Commodore got a failing chip foundry at a bargain price, not a cutting edge one (and it served them well, as they went on to produce the best selling desktop computer of all time!).

As for producing a 16 bit version of the 6502 that IBM would use in the PC, you're dreaming. The 8088 was the obvious choice because it provided an upgrade path from 'serious' 8080 based computers, while even Apple shifted to 68000 because the 6502 was too limiting. WDC did later produce a '16 bit' 6502 but it too was limiting and never became popular, and other manufacturers who tried to produce 16 bit versions of their 8 bit CPUs failed. The idea that Commodore could have been the one to beat them (if only they had made the 'right' decision) is laughable.

It's always the same refrain - if only Commodore had done this or that they they wouldn't have crashed, leaving us feeling inadequate! But other computer manufacturers who did do those things didn't survive either, and Commodore outlasted many of them.