Why did DOS/86 overtake CP/M-z80 ?

[Eudimorphodon]

Can you express the number of major releases for the C64 the 1990's as a percentage of all major releases for the C64? If it's a big number like 10%, I'll agree the C64 was still in its prime. If it's a small number, say 5%, then I'd say it's still doing amazingly well for its age. If it's around 1-2% then I think we can conclude that the C64 was in its dying years. That's why I provided the releases as a percentage, not as an outlier. I'll accept ports in that figure also. Seeing the graph gives a good feeling for the platform's actual life.

Maybe I’m blind, but where exactly is this graph you’re demanding I produce for the Commodore 64 in that article about the PC-88? Moby Games lists 319 games released for the C64 in 1990, verses 909 in 1984, 622 in 1987, and 107 in 1993. So, sure, obviously it’s tailing off, but that 319 figure is still a 1/3rd of the output for the machine the year it actually single-handedly outsold *all* PC clones in the United States.

Moby Games lists 556 *total* titles for the PC-88 platform, 57 of which are from 1990. That number drops to 20 the next year, and 1992 is the last year listed before the 21st century with *3* releases. Coincidentally peak year for the PC-88 was also 1984, with 92 releases.

These figures aren’t helping your case much, at all. All they show is the PC-88, clearly a machine with something of a cult following based on the relatively small total software base, had a somewhat flatter curve through the very end of its life than the *extremely* mass-market/mainstream C64 did. These are not surprising results. They in fact make it crystal clear that 1990 was the PC-88’s last “good“ year, not that it was in anyway viable into the 1990’s for *anything*.

I was still playing a huge pile of pirated Atari 800 software into the 1990’s despite having a 486 in the house, why are you assuming these hardcore porn gamers (a fair number of the late PC-88 games are NSFW) used their obsolete Z80 machine as their real computer based on *game sales*? *You* have to make that case if you want to claim the world made some kind of mistake abandoning CP/M “too soon”.

 

[vwestlife]

Actual worldwide sales figures for the C64, from Commodore's own annual reports, by fiscal year:

'83: 500k
'84: 2.5 million
'85: 2 million
'86: 1.9 million (includes C128)
'87: 1.1 million (includes C128)
'88: 1 million (includes C128)
'89: 1 million (includes C128)
'90: 700k
'91: 800k
'92: 650k
'93: 200k

So yes, the C64 was still selling well into the early '90s, but mostly as an alternative to a video game console, and as a first computer for Eastern Europe and other developing markets.

 

[Eudimorphodon]

So yes, the C64 was still selling well into the early '90s, but mostly as an alternative to a video game console, and as a first computer for Eastern Europe and other developing markets.

IE, exactly the same reasons ZX Spectrum clones and a few niche Japanese machines also staggered along a few years past the fall of the Berlin Wall.

The thing that’s really clear here is just how irrelevant “CP/M”, the operating system, was in this picture. The C64 didn’t even have a CPU that could run it, the vast majority of ZX Spectrum machines never even had disk drives, and the deciding factors in what’s left was always the machine’s cost and/or suitability for games, not whether or not it ran Wordstar. It’s absurd that we’re diving into how many 8-bit Hentai games were written for obscure JDM computers looking for reasons that an operating system that barely understands teletypewriters should/could have stayed relevant longer.

 

[danielbooneamerica]

TRSDOS was destroying everything. No OS could compete with TRSDOS.

A rumor TRSDOS was being developed to run on IBM mainframes pushed IBM to enter PC market. Radio Shack refused to license TRSDOS to IBM thus development of MSDOS.

After years of corporate wars between Radio Shack and IBM, TRSDOS was etching out MVS, TSO, VTAM and many other IBM products. IBM then began a backdoor deal with Mr. Roach then president of Tandy.

Due to billions being passed in background, Tandy was slowly let to slip in market. TRSDOS for mainframes was put on back burner and Roach retired with billions in a Swiss bank account.

Today TRSDOS for IBM mainframes is used only for national defense. If you ask defense department, IBM or any leading tech firm they willl tell you they never heard of this.

But that is just a cover story.

;-)

 

[Agent Orange]

Toy systems such as the Model One or the Sinclair didn't have a chance against the IBM PC. Expensive S-100 systems such as my 6Mhz Z80/twin 8" DSDD floppy system plus a 1/4Meg MDISK ran circles around any early IBM PC, but that wasn't the issue. Tandy had their Model 2 and follow on systems, but priced them to the moon. IMHO, the lack of a standard disk format (or even diskette size) plus no std. video format more than anything killed off CP/M. If one could just stop at a local computer store and pick up software and drop it into a floppy drive and have it run on multiple brands of computers I think that would have extended CP/M's life by a few years.

The closest std. disk format, I'll guess, was the 8" IBM single sided format that Digital Research and others used. Or perhaps the Osborne format(s)? Even Tandy, at first, had different formats on their 5.25" systems depending upon what floppy OS or supplier created one for the Model One. Everyone was dependent on either the maker of their system or the software developer to supply software on a disk that would run on their system. Need some new program, please please Mr. Software company pleease provide it on the special format my machine runs. God Forbid it was a hard sector format! Price? Whatever the market would bear! Then spend an hour or two configuring your terminal's needs to match the functions of the program. Remember keyboard overlays? Then there was the issue so many times that if you had a friend with an Osborne running the same programs and you had a Epson QX-10 or a Morrow (or or or) you'd have no luck in reading each others diskettes. WTF would you do with an 8" beast brought home from work? So the answer often was transfer with a 300 BPS modem! Yikes!!! Tie up your phone line for hours too. Or upgrade to 1,200? Now just 30 minutes at long distance rates. Wow! Remember. often the next town was a long distance phone call back then!

Or switch to a PC and stop at the local computer shop and go home with a cheap as chips new software package. Let's face it, much more often just go to your friend's house and get a bootleg copy for free. Disks swapped back and forth and you could mail them for next to nothing too. Perhaps that's close to the real answer. Bootlegged software killed CP/M.

I don't agree with your assessment of the Model 1 as "toy" as it introduced many to BASIC programming for under $700, and we used it in our lab to run actual formulas via the tape input route. Again, the Model II was business machine and a low cost option compared to IBM. It was never intended for home computing and most, if not all of the off the shelf software was business oriented,

 

[Chuck(G)]

@danielbooneamerica -- ROFL!

 

[Eudimorphodon]

I don't agree with your assessment of the Model 1 as "toy" as it introduced many to BASIC programming for under $700, and we used it in our lab to run actual formulas via the tape input route.

The Model I outsold everything else on the market in the US for almost 4 straight years and was, when fully expanded, about as capable as any other first gen personal computer so, no, clearly not a toy. If Tandy hadn’t ruined a lot of what I perceive at least as the strengths of the machine (extremely low cost of entry relative most of its competition and modular construction) with the Model III it may well have had a couple more good years… but sure, obviously lacking some major upgrades a $399 computer from 1977 is going to look pretty crap compared to a $1,565 computer from 1981. (Comparing base prices minus cassette recorder and monitor here, obviously.)

 

[Chuck(G)]

As far as the Z8000 goes, in my view, it was a better design than the 8086--and could easily be incorporated in embedded designs with the Z8002. But in the 1980s, Zilog was a chancy proposition; no one knew what Exxon was going to do with it. Witness the fate of Quip, Vydec, Xentex, Qyx, etc. Qume and Zilog managed to escape, eventually; the others not so much.

Now, if you were an OEM looking at MPUs, who would you choose? Intel was a pretty safe bet, with quite a number of alternate sources. Motorola had the edge on a technically-advanced MPU backed by a solid company with a history, but supply issues dogged it in the first few years of the 68K.

Product decisions are not made solely on technical grounds.

 

[krebizfan]

The Z8000 managed to entice many companies to creates schematics that would use it. Zilog didn't have the ability to produce sufficient engineering samples to turn those designs into working prototype hardware. Something existing is a very relevant technical issue.

Tandy was a bit strange in that they tried to create systems with a number of alternate CPUs before deigning to go with the 8086/8088.

 

[danielbooneamerica]

I agree model 1 was a toy.

my 1st wife yelled at me endlessly while using a model 1.........he is playing with his computer again.

That had to have happened at least a million times and she was a pretty smart girl, so it must have been a toy.

 

[danielbooneamerica]

I agree model 1 was a toy.

my 1st wife yelled at me endlessly while using a model 1.........he is playing with his computer again.

That had to have happened at least a million times and she was a pretty smart girl, so it must have been a toy.

No its not a toy. In 1981-82 at NAS Kingsville, Fred Haass used a model 1 to help assemble siimulation programs.
We read most of program in using punch cards (100% assembler).
These sims used two Sigma 5 CPUs connected via common memory.

We assembled from punch cards and wrote object out to paper tape.
Back then we had to make 2 passes manually over source to resolve forward references.

After assembly, obleject code on paper tape was read and object code from previous assembled libraries was linked (all on paper tape).

But memory was so tight we really didnt build or sort memory tables. It is super useful in simulation to know memory location holding a value or flag value.

We used symbol tables feed into a model 1 where we could then sort and massage into output for a printer (and printed using a Radio Shack line printer).

Fred also printed loan amortizations for his wifes real estate business.
NAS simulation office had a rack with tapes for model 1. We kept program writen to sort symbols on these tapes.
Fred labeled tape for his wife business AMY (amortization).
When higher ups from Washington was on a tour we told them AMY stood for a program that tested data on Analog Mechanical Y axis.
Heck they didnt know any better and said good job.

Point of all this ramble.....Fred Model 1 did useful work making sims fly and also for his wifes real estate business, so no it was not a toy.

BTW.....those sims were for Navy A4 jets......you may have seen them, flown by Blue Angels at that time.....we were very proud to be involved.

Today those 2 Sigmas are replaced with 25 or so power servers....not sure its an improvement but it is cheaper.

 

[Chuck(G)]

The Z8000 managed to entice many companies to creates schematics that would use it. Zilog didn't have the ability to produce sufficient engineering samples to turn those designs into working prototype hardware. Something existing is a very relevant technical issue.

AMD was a Z8000 licensee, even going in a joint venture with Siemens to form AMC (not the car manufacturer). The venture didn't last long. I've never seen an AMD-branded Z8000 MPU of any stripe, though they probably exist.

 

[DeltaDon]

Clearly, my use of "toy" was not seen as being tongue in cheek. Many model 1's were used for small businesses, but not many Fortune 500 companies had them. Or if they did they weren't out in front of the office for everyone to see. I'm sure a lots of engineers had them at home, and maybe at work too, and used them until the silver gray paint wore off. Few in any ever ran CP/M. Ditto with the Sinclair computers. Maybe they used Z80's, but were not running CP/M. IIRC, the company I worked for had Lanier word processors in front of every secretary and the engineers were shared terminals tied into the company mini computer until PC's were on most every desk. They tried to make a S-100 machine control running with a single 8080 CPU that was a total disaster - the poor overworked beast couldn't even allow input when trying to run a machine. But hey, they were an analog electronics company up until that point.

My own CP/M computer collection had and has machines that used various 5.25" diskette formats, several 8" custom formats (plus all able to read 8" IBM format) and one computer with a 3.5" floppy (The NEC Starlet laptop). Swap diskettes, not in this lifetime. I also remember back in the early '80's, trying to get CP/M FOG (remember FogHorn?) programs from the big discount software store in Orange County, CA - 5.25 Osborne diskettes or go home. But lots of public domain stuff in IBM PC format for 8088 stuff. I did manage to buy DBase II on Epson format very cheap and got a free "update" from Aston-Tate on 8" diskettes.

 

[Agent Orange]

No its not a toy. In 1981-82 at NAS Kingsville, Fred Haass used a model 1 to help assemble siimulation programs.
We read most of program in using punch cards (100% assembler).
These sims used two Sigma 5 CPUs connected via common memory.

We assembled from punch cards and wrote object out to paper tape.
Back then we had to make 2 passes manually over source to resolve forward references.

After assembly, obleject code on paper tape was read and object code from previous assembled libraries was linked (all on paper tape).

But memory was so tight we really didnt build or sort memory tables. It is super useful in simulation to know memory location holding a value or flag value.

We used symbol tables feed into a model 1 where we could then sort and massage into output for a printer (and printed using a Radio Shack line printer).

Fred also printed loan amortizations for his wifes real estate business.
NAS simulation office had a rack with tapes for model 1. We kept program writen to sort symbols on these tapes.
Fred labeled tape for his wife business AMY (amortization).
When higher ups from Washington was on a tour we told them AMY stood for a program that tested data on Analog Mechanical Y axis.
Heck they didnt know any better and said good job.

Point of all this ramble.....Fred Model 1 did useful work making sims fly and also for his wifes real estate business, so no it was not a toy.

BTW.....those sims were for Navy A4 jets......you may have seen them, flown by Blue Angels at that time.....we were very proud to be involved.

Today those 2 Sigmas are replaced with 25 or so power servers....not sure its an improvement but it is cheaper.

When I was the there, 1958-60, 1st stop out of boot, it was NAAS Kingsville and north field was flying the F9F Panthers and south field had the S2F Trackers and the T-28. We had punch cards and the only computer was a quasi mainframe at Main Side - NAS Corpus Christi Supply Center (3rd floor).

 

[cj7hawk]

Maybe I’m blind, but where exactly is this graph you’re demanding I produce for the Commodore 64 in that article about the PC-88? Moby Games lists 319 games released for the C64 in 1990, verses 909 in 1984, 622 in 1987, and 107 in 1993. So, sure, obviously it’s tailing off, but that 319 figure is still a 1/3rd of the output for the machine the year it actually single-handedly outsold *all* PC clones in the United States.

Moby Games lists 556 *total* titles for the PC-88 platform, 57 of which are from 1990. That number drops to 20 the next year, and 1992 is the last year listed before the 21st century with *3* releases. Coincidentally peak year for the PC-88 was also 1984, with 92 releases.

These figures aren’t helping your case much, at all. All they show is the PC-88, clearly a machine with something of a cult following based on the relatively small total software base, had a somewhat flatter curve through the very end of its life than the *extremely* mass-market/mainstream C64 did. These are not surprising results. They in fact make it crystal clear that 1990 was the PC-88’s last “good“ year, not that it was in anyway viable into the 1990’s for *anything*.

Yep, that pretty much coincides with what I said. The PC-98 came out around the same time as the PC-88 but it wasn't until around 1990 that it replaced the PC-88... And we can reasonably say it was doing well until the end of the 80s, then quickly died out.

You mentioned some longer timeframes for the C64 and I was wondering if you knew what they were as a percentage so we can understand the significance of those numbers with relation to the total number of games released for the C64. If you don't know the figures, a guess is fine. It will still give me more context into what you're saying.

I was still playing a huge pile of pirated Atari 800 software into the 1990’s despite having a 486 in the house, why are you assuming these hardcore porn gamers (a fair number of the late PC-88 games are NSFW) used their obsolete Z80 machine as their real computer based on *game sales*? *You* have to make that case if you want to claim the world made some kind of mistake abandoning CP/M “too soon”.

I never claimed the world abandoned CP/M too soon. CP/M was never supported the way Microsoft supported it's own OS products... Microsoft killed CP/M. That much is absolutely clear.

I was just curious as to why, in the US market, it never even put up a fight? (not including DR-DOS) - It certainly persisted elsewhere for longer, as demonstrated.

Though if your answer is "because we never got porn on CP/M systems in the US" well, it's a different perspective - I'll grant you that -

Maybe if "Deep ASCII" was available back in the 80's it might have been a different story?

 

[danielbooneamerica]

When I was the there, 1958-60, 1st stop out of boot, it was NAAS Kingsville and north field was flying the F9F Panthers and south field had the S2F Trackers and the T-28. We had punch cards and the only computer was a quasi mainframe at Main Side - NAS Corpus Christi Supply Center (3rd floor).

you were a TD?

 

[cj7hawk]

Actual worldwide sales figures for the C64, from Commodore's own annual reports, by fiscal year:

'83: 500k
'84: 2.5 million
'85: 2 million
'86: 1.9 million (includes C128)
'87: 1.1 million (includes C128)
'88: 1 million (includes C128)
'89: 1 million (includes C128)
'90: 700k
'91: 800k
'92: 650k
'93: 200k

So yes, the C64 was still selling well into the early '90s, but mostly as an alternative to a video game console, and as a first computer for Eastern Europe and other developing markets.

Thanks for that - it looks like '92 was the last good year for the C64. It really hung in there too.

 

[cj7hawk]

Whether or not a computer was a "Toy" merely relates to how it is used. Even the smaller systems like zx80s got used for some serious work at times... IIRC, there was an example of the zx81 being used to simulate running of a nuclear reactor floating around the web somewhere. I know of examples of one company using them to create reticulation controllers and several of them were running until recently here in Australia even.

And Basic, much maligned today, really was a great language... I never learned to love any language as much.

But I do recall there was a longing in the Sinclair camp for our machines to be able to run CP/M - there was certainly enough technical experience around, but the lack of RAM in the zero page put an end to those ideas, and if you got past that point, the video ram was your next issue.

It certainly would have been possible to achieve though... which really makes me wonder why no one ever did it, but I guess it's one of those "no one really thought much about doing it at the time outside of talking about it" situations.

In this respect, many people who owned Spectrums thought of them as "toys" and knew the step between "toy" and "business computer" was little more than the OS and some memory management away.

 

[krebizfan]

But I do recall there was a longing in the Sinclair camp for our machines to be able to run CP/M - there was certainly enough technical experience around, but the lack of RAM in the zero page put an end to those ideas, and if you got past that point, the video ram was your next issue.

Amstrad marketing would have tried to remind people that the Spectrum +3 ran CP/M.

I have looked at a number of the clever Z-80 systems that supported multiple memory maps and am convinced that it cost more to work around the limitations of the Z-80 than was saved by not using an 8088.

 

[cj7hawk]

Amstrad marketing would have tried to remind people that the Spectrum +3 ran CP/M.

I have looked at a number of the clever Z-80 systems that supported multiple memory maps and am convinced that it cost more to work around the limitations of the Z-80 than was saved by not using an 8088.

From a cost-only perspective, this is true. However I did find that by using BDOS as a Memory Manager, and treating the entire RAM space as a DISK, you gain enough advantages using CP/M that the 64K window that the CPU has into memory is no longer an issue.

I think the thing that impressed me most when using the defacto BDOS as a MMU was that you can see your entire memory laid out in the file allocations, so I tried logging into the disk assigned to the MMU and typing "DIR" and all the processes came up... Using other tools like STAT, I could see what each process/driver etc, was using. Some of the better tools even told me what memory locations they were consuming... And of course, any application can use the BDOS then to access the extended memory directly as an extended data space.

It may not be as fast as directly paging it in, but it addresses all of the shortcomings I found existed with most paged memory solutions.

I did try CP/M+ for the ZX SPectrum +3... It works pretty well. The biggest issue I have with it is a lack of 80 column display. That's what got me interested in CP/M in the first place.