We know some numbers for the 6502 based PCs:
1) for the Commodore 64 - about 17 million;
2) for the Commodore VIC-20 - above 1 million;
3) for the Atari 400/800 - about 4 million;
4) for the BBC Micro - above 1.5 million;
5) for the Commodore 264 family - above 1 million;
6) for the Apple II - about 5 million.

I can estimate the number for the Commodore PET and other the 6502 based PCs as about 1 million.

So we have about 30 million units sold.

We know also some numbers for the Z80 based PCs:
1) for the MSX - about 8 millions;
2) for the Amstrad CPC and PCW - more than 11 millions;
3) for Tandy TRS-80 - more than 2 millions;
4) ZX Spectrum with clones - less than 10 millions.

I can also estimate that the number for other various Z80 computers sold is less than 1 million. I am not sure about the Tandy TRS-80. Maybe the total number for model 1, 2, 3 and 4 is larger.

We have got about 32 million units sold.

We also can add information about game consoles.

For the 6502 based:
1) the NES - about 62 million;
2) the Atari 2600 - about 35 million;
3) the Super NES - about 50 million (the 65816 is compatible with the 6502).

So we have got more than 147 million.

For the Z80 based:
1) the Master System - about 13 million;
2) the Game Gear - about 11 million;
3) the Mega Drive - about 33 million;
4) the ColecoVision - above 2 million;.

So we have got a number about 60 million.

I can get a conclusion that the number of the 6502 systems were larger than the z80 systems. However the Z80 was used in very popular the TI-83 calculators and I don't know any popular the 6502 based calculators...

It would be interesting if anybody can give more information about subject or make corrections to information given afore.

Thank you.

_________________
my blog about processors

Game Boy used something related to the Z80, so that's another 118 million if we allow it.

The notebooks NC100, NC200 and Z88 were all Z80 based.

I believe the Amiga at least used a 6502-like processor in the keyboard, and Commodore liked to use 6502-like processors in their disk drives - do we count those??

HP made a number of calculators with a 6502-like processor in them, including at least the HP 12C Platinum (4 models), the HP 12C Prestige, the HP 35s, the 9g, the ever-popular 17bII+. I suspect TI-83 might tip the balance though, as they own the US school market, AIUI.

Very many toys use 6502-like cores, including at one time the very popular photo keyrings. But these are not devices for calculation or computation or gaming.

There were a number of S-100 bus computers using the Z80.
Other systems including Osborne, Kaypro, Xerox.
Don't forget that the Commodore 128 includes a Z80.

Ohio Scientific made 6502 computers.
There were a number of Apple II clones, Franklin being among the most well known.

Add-on cards or cartridges to run CP/M on otherwise incompatible computers, most of those included a Z80.

While not computing platforms, many arcade and pinball machines used either a 6502 or Z80.

From hackaday:
"you would not believe how many toys are still shipping with a 6502-ish core somewhere inside"

(Bearing in mind the 6502 has about a quarter the transistors of the Z80, wherever it is sufficient it is surely going to be cheaper.)

Also Z80-based: the Sinclair ZX81 and the Timex versions thereof, the Timex Sinclair 1000 and Timex Sinclair 1500. Also the ZX81 predecessor, the ZX80.

-- Jeff

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

Don't forget these...

Atari 5200
Atari Lynx
Oric Atmos
Acorn

The 6502 was used in every Commodore Disk drive outside of the Amiga, so add many millions of those too. The disk drive was a computer unto itself with RAM and ROM and a CPU. If you count the z80 in the Genesis/Megadrive, you should count the 10 or more million Commodore disk drives out there

Add 17 million plus for the 6502 tamagotchi, one of many toys mentioned above.

If anyone missed it, the 6502 source code from the tamagotchi is available for download

Edit: make that 82 million as of 2017, I can't copy-paste effectively.

At one point I came across a dot-matrix printer whose serial interface board included a full-blown Z80, in addition to the processor of the printer itself (which had the relatively complex task of turning ESC/P style commands into motor and solenoid signals).

Embedded CPUs, when cheap enough, get literally everywhere - a fact which ARM has exploited pretty ruthlessly.

Thank to all for some pieces of interesting information. It looks that the 6502 systems were prevailing over the z80 until the 90s. It really fascinates me very much. Zilog had two new fabs in the 70s, large funding from Exxon, even personal support from Bill Gates for the Z80, ... The 6502 had almost nothing behind it but its quality.
I think it rather senseless to count the 6502 or other processors in cases where they were used as controllers. They are too numerous. Chuck Peddle in his Oral history mentioned his ultra-fast SSD based on ten 6502s. Has it been produced somewhere?

_________________
my blog about processors

Gameboy uses rather the enhanced 8080. Its CPU doesn't have index registers and many other the Z80 features.

_________________
my blog about processors

IMHO the Z80 in the C128 was rather dormant. Its effective frequency is only about 1.6 MHz, it can't effectively use VIC or SID, there was no almost any support for software for it. People bought the C128 as an upgraded C64 but got exactly the C64 and several almost unsupported hardware options.

_________________
my blog about processors

Maybe. Every time you powered on the 128, the Z80 was in charge until it finished polling the expansion ports, etc before handing control back over to the 8502.

Pretty much every single device on the 128 was accessible to both the 8502 and the Z80. The *ONLY* exception to that is tape and the CAPS key which were tied to the I/O pins on the 8502.

You can most definitely access the VIC and the SID from the Z80 directly (you do *NOT* have to go through he 8502 to do it).

Very few ever did it and the BDOS from Commodore themselves actually just flipped control of system from the Z80 back to the 8502 for nearly all I/O, but it most definitely can be done.

The *ONLY* reason I know this is one of the early build environments that I worked on which used CPM on a 128 for cross development work for 65xx, 68xx and Z80 actually used the SID as an audio notification as to when to change disks (the compilers, assemblers and linkers had to be continually paged in from disk while it did it's work so you would just start it running and it produced a "perpetual chime/tune" on a 128 when you had to change disks).

Adding a REU (when they eventually become available here in AU) was a godsend and increased the speed by a factor of about 100x when it ran (and also did away the horribly annoying disk changing shuffle).

I should be able to remember the name of the package and the supplier here in AU as they also sold a bunch of modifications for the 128 as well to "improve" the system but for the life of me, I can't .... 30+ years ago and a whole lot of alcohol and my memory is shot ...

There’s also Rockwell modems. I don’t know how many were sold, but as the Internet was getting started they dominated the market. Each modem had a 6502 in it and the central site modem was also 6502 based for much of the time until they moved to ARM. Maybe a few 10’s of millions of 6502’s were used here too.

I wrote exactly about the same things but experts who tried to use the SID and VIC with the Z80 reported that it is possible only for very simple things because of timing issues and other problems. Do you know any demo for the Z80 using the SID or VIC? IMHO the C128 was good only as the C64. Basic 7 was slower than slow Basic 2. No support for the VDC was provided. The C128 was really a great marketing success for Commodore but IMHO they rather deceived ppl, they could make a much better computer... IMHO knowledgeable people should have bought the C64 and Amstrad CPC6128 instead of the C128 to get much better option. Commodore even artificially cut MMU to support only 128 KB.

_________________
my blog about processors

Might be right but I have no idea. I was only ever considered myself a "user" of CP/M - I never actively developed in it (or for it). Yes, I wrote some tools to ease the pain with the cross-build process that we used for supporting various SBC control boards that I supported, and yes I could get it to make a beep and and whistle and those *DID* use the SID and played a simple chord but that was just for notification - nothing more. I will be honest and say that I didn't even write the routines that did it - I just called a utility from a SUB script to do it. From a technical perspective, there is *NO* reason as to why you can't access any I/O chip (sound, video, etc) from the Z80. They share the same bus. If there are timing issues, then it is probably down to the way that the Z80 itself works in conjunction with it's implementation in the 128.

You have sparked my curiosity though so I'll have to drag my C128DCR back out of the attic and fix it so that I can actually experiment with this (it's dead at the moment with a bad DRAM chip and I've just been completely slack about getting around to fixing it - been on my "I'll get around to fixing that one day" list of things for a few years now ...).


Note sure I agree with this. It was both the slowest of the CBM basics, and also the fastest depending on if you were in "FAST" mode or not.

Part of the problem/benefit was that BASIC variables and program code were stored in a separate banks so BASIC was forced to switch back and forth continuously.

The BASIC7 was also significantly more advanced and therefore because it supported so many more features was naturally going to be slower. That said, I'll take BASIC7 over BASIC2 any day of the week, even if it was ~ 15% slower.

Pretty much. The VDC was an absolute b*******d of a chip. I think graphics was an afterthought and the VDC was only ever considered for text only.


Ok, this one I have to laugh at. Commodore marketing was abysmally bad at the best of times (at least it was here in AU). One of my mentors (he worked at Commodore AU) used to state that Commodore marketing weren't competent to to sell water to a man dying of thirst in the desert. Based on my own experience at the time dealing with the chip division (we used the MOS 6502s and 6526s in some of the PLC boards) that is 100% accurate.


As far as CP/M was concerned - maybe. I never had a CPC so I just don't know. I had a 128 because it had a Z80 in it *AND* it had a 6502 "like" processor in it, and was a semi-reasonable build system (albeit a bit slow). Being able to have just the one computer and not three was pretty important in the day.

Ok if I'm to believe Bil (Herd) the MMU came straight from the LCD computer, and number of the "limitations" were inherited from that design. The lack of the 256Kb of RAM (one of my great bugbears) was a side effect of limitation of the LCD project (the 8722 was only half working when it came from the LCD machine and it needed a fair bit of additional work for the 128). How true that is, we'll probably never know as Dave (DiOrio) is now gone.