3

What's a good way to estimate the total number of 6502 or 8051 chips (or workalikes, pin-compatible or not, but not including software or FPGA emulators) ever manufactured or sold?

Added:

(To clarify, for the 8051, I'm asking about a estimate regarding the number integrated circuit devices that execute the original basic 8051 ISA directly using hardwired transistors dedicated at fabrication time for executing the 8051 ISA, or an inclusive proper superset. Similar question for the 6502 ISA. Disregard differences in the originally documented as unimplemented instruction space.)

  • 3
    Note that the 8051 is still in production: ATMEL sells pin-compatible variants as well as non-pin-compatible variants that have USB and other "new" features. All using Flash ROM instead of EPROM. – Martin Rosenau Feb 22 '19 at 8:54
  • 1
    I'm wondering what is 'software' or 'FPGA' emulators for you? Do you consider netlist-level simulator of 6502 at visual6502.org as 'software emulator' as well, and if yes, why? Do you consider gate-level design of 6502, translated into HDL and then compiled to work inside an FPGA, as 'fpga emulator' as well? – lvd Feb 22 '19 at 13:37
  • 2
    @Raffzahn My point is: Today there are still 100% pin-compatible variants of the 8051. And even the other 8051-compatible chips do the same an original 8051 did: They work as microntroller (with on-chip GPIO, UARTs, timers, memory ...). The original 6502 was a pure CPU without any on-chip peripherals or memory. Now the question is: Does a 6502-based microcontroller (with on-chip peripherals and memory) count as "6502" or not? This question has to be answered before the original question can be answered. – Martin Rosenau Feb 22 '19 at 16:36
  • 1
    @Raffzahn I already wrote that the 65C02 is still produced, but I doubt that a large number is sold. My point is: Because the 6502 has no GPIO pins (unlike the 6510), the only function an original 6502 is to exchange data using an external data bus. A Renesas 380L seems to be a 6502-based microcontroller without external data bus. This means that the only function of a 6502 (exchanging data with an external data bus) is not supported by the 380L. Is the 380L a "work-alike" device? On the other hand you could argue that the 380L contains a 6502 internally... – Martin Rosenau Feb 22 '19 at 20:02
  • 1
    Just wave your arms and say "A lot. A whole lot." – user12 Feb 22 '19 at 20:05
11

Almost impossible to tell. Both CPUs have been sold by quite a lot of manufacturers in many variations - including knockoffs modified in some way to avoid royalties.

One hint might be the claim of 5 to 10 billion (*1) on the WDC site:

The legendary 65xx brand microprocessors with both 8-bit and 8/16-bit ISA’s keep cranking out the unit volumes in ASIC and standard microcontroller forms supplied by WDC and WDC’s licensees. Annual volumes in the hundreds (100’s) of millions of units keep adding in a significant way to the estimated shipped volumes of five (5) to ten (10) billion units. With 200MHz+ 8-bit W65C02S and 100MHz+ 8/16-bit W65C816S processors coming on line in ASIC and FPGA forms, we see these annual volumes continuing for a long, long time.

Since WDC is nowadays the major licensor (*2) and still can only give it with such a huge uncertainty, I doubt anyone will be able to give a better one.

I think it's safe to assume this as well for Intel ... except, here the numbers may be way higher. 8051 (and derivatives) have been in heavy use all around the world by major suppliers.


*1 - US Billion, Milliards to others, 10^9 for serious minded.

*2 - MOS/Commodore gone for good, most others as well, there is still RENESAS and Rockwell, except I'm not sure if they license the core at all.

|improve this answer|||||
  • There are now several 8051 cores: basic cores with 12 clocks per cycle, derivatives with 6 clocks per instruction and 2 clocks (all those are much alike since they keep count of cycles per every instruction), and as well, cores with 4 clocks per cycle and 1 clock per cycle, where cycle counts are broken. Then again, while still being compatible on ISA level, they all have completely different designs, made presumably from scratch every time. Shouldn't they again count as 'FPGA/HDL/silicon emulators'? :) – lvd Feb 22 '19 at 19:31
  • And just to make my point clear: if one reverse-engineers some retro CPU core down to transistor level, recreates gates and generates HDL that behaves exactly the same (with the possible exception of too-much-analog things like unstable undocumented instructions in 6502), this is what I wouldn't count as an emulator, be is new silicon, FPGA firmware or netlist simulator running on PC. Everything else is more or less an "emulator", including compatible cores made from scratch. – lvd Feb 22 '19 at 19:34
  • Why did you roll back the edit? Do you have any specific reason to have a screenshot instead of textual citate? – peterh - Reinstate Monica Mar 6 '19 at 13:10
  • 2
    @Raffzahn: I don't disagree with you. Just trying to remind people. In practice roughly half of images posted get alt text. I do add text to images when I notice that it is missing but nobody can catch them all. – Chenmunka Mar 6 '19 at 13:51
  • 1
    For having seen the source code of Rockwell based modem chipset (remember the Hayes POTS modems?). They all used a 6502 based microcontrollers. They were also running at insane frequencies. The model we had signed the NDA for (was in the 33600 bps generation of modems) ran at 70Mhz if I remember correctly. They were coupled with a DSP to decode the line signal. The µC handled the protocol (HDLC synchronous) and the interpreter (the famous AT+ commands). – Patrick Schlüter Mar 20 '19 at 9:15
3

Regarding the 8051:

  • One in most OSD-capable CRT monitors or television sets ever sold.
  • One in most pre-DVD CDROM drives ever sold.
  • One in most pre-2000s hard drives.
  • A predecessor of the 8051 (the 8042. An 8048 derivative. The 8048 is NOT an 8051 per se, the 8051 was developed as a solution to do what the 8048/-42 could not do well and is NOT an extended 8048/8042) hardwired into most any PC chipset.
  • One in most PC keyboards.
  • An 8051 or 8048 in many household appliances that were button-and-LED controlled.
  • One 8051 or 8048 in most VCRs and VFD-interfaced HiFi devices.
  • One 8051 (or 80196! Another intel MCU, not an 8051), often the SAB80C515 or SAB80C517, in quite some early car ECUs.
  • Probably a couple of them around even a modern car.
  • One 8051 in a typical electronic central heating controller.
  • Often, one 8051 in a telephone with a non-graphic LCD display.
  • In an SD card or USB stick, often a supercharged 1-cycle 8051 core in the controller chip (likely hardwired for cost efficiency).

Generally, anything where a mechanical control solution (that isn't math heavy) is replaced by digital control, but without a graphic LCD (handling display fonts on an base, 12-cycle 8051 doesn't leave much capacity for anything else. Especially all the bit-wrangling to drive displays that want their data as 8-bit pieces of a row!), suspect an 8051 at work.

Often not 8051: film cameras (very custom stuff), toasters (4 bit cores), electronic door locks (PIC), calculators, simple digital watches (hardwired), printers (80188, ARM), optical mice (68HCxxx)...

|improve this answer|||||
  • 1
    Do you mean 12-cycle (vs 16-cycle) 8051? – supercat Feb 27 '19 at 22:37
  • 1
    Nice writeup, except that the 8048 (etc.) is not part of the MCS-51 family. While the MCS-51 was developed as a replacement for the MCS-48, it is not plug compatible - not even binary. Compatimbility exists only on source code level and only with restrictions. So it's more like 8088 and 8080. You may want to remove these references for a clear picture. – Raffzahn Feb 27 '19 at 22:55
  • Oh, and similar the 80196. It's a member of the MCS-96 family (8061, 8096, ...), which is similar to the 8051, but an independant development as well. Again only partiality source code compatible. – Raffzahn Feb 28 '19 at 0:11
  • 2
    Clarified. I think mentioning the MCS-48 and MCS-96 is important in that context, though, since these share some of the markets mentioned... And actually, the 8048 is as ugly as a PIC in architecture :) – rackandboneman Feb 28 '19 at 8:31
  • 1
    Wouldn't underestimate keyboards - given that many of us, in the corporate world, do treat them as disposables these days, probably wearing out 3 to 5 of them over the lifetime of a desktop PC. Also, I suspect you have a lot of MCS-48 in the dishwasher and washing machine world too... – rackandboneman Feb 28 '19 at 9:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.