Overview of 6502-like cores, hard, soft, partial, overblown

[kakemoms]

As you may notice I keep updating the list with entries from the replies. I want to limit the list with respect to the number of entries. I.e. we can´t add 600 different mask-programmable Mitsubishi controllers, plus there are far too many other kinds of micro controllers with 6502-like cores that needs to be mask programmed (thus are unusable for most of us). For historical interest I can add some, but it will be limited.

Rockwell had some versions (of the kind with mask programmed roms) that can be used with external PROMS, even in the NMOS era. Such are more interesting to the general user as they can be programmed at home/in office and thus can be used for a small number of prototypes. So I generally would think its better to add (a few) of the kind that can be used with an external ROM.

Its easier to add things if you can link to a data sheet. Some old IC´s tend to be limited to prototypes and datasheets are non-existent. So I will limit the list to whatever one can get a datasheet on.

Anyway, happy to add more!

[Cray Ze]

I don't think anyone has suggested to add the 600 Mitsubishi micro controllers (the thread is about cores isn't it?), which are not all mask-programmable by the way, some are external ROM, some are Flash. Many also have a full external address and data bus. There are probably five or six cores spread across the range, most of which can run 6502 code directly.

I hadn't finished with my post near the top of this thread yet, there's a lot of data and not always a lot of time, but if there's not interest, well, that's easier for me.

[BigEd]

I'm interested in different cores, in the sense of different subsets or supersets of the original 6502.

But the Rockwell chip with an on-chip Forth ROM is also interesting. If it so happens that it has one or two extra instructions, it qualifies on that merit.

Your post here
viewtopic.php?p=56262#p56262
is certainly interesting too, Cray Ze. If you do update it, do let us know by posting a bump. If there's not a complete intersection of interests, there's always room for another thread!

[Cray Ze]

I'll be updating the post soon. They are interesting from a hobbyist point of view where obsolete parts don't really matter. They are also still easy enough to find.

I just bought 4 x M37700SAFP to have a play with.
http://www.ebay.com.au/itm/x/371493459897

These are ROMless, requiring external ROM, and have a 16MB address space.
A datasheet can be found here, pages 40-43 give info on microprocessor mode.
http://www.datasheetspdf.com/PDF/M37700SAFP/546510/1

[RalfK]

Some datasheets can be found here
The R6511AQ: http://6502.org/documents/datasheets/ro ... 500-13.pdf
More: http://6502.org/documents/datasheets/rockwell/

The R6511AQ is elder than 1988. The datasheet mentioned abbove is a Rev.5 from 1987. I open my old Rockwell "bible": R6511AQ Rev.4 October 1984. The Applebus board was developed in 1984.

This chip is able to address 64kB external RAM. 8 IRQs are generated internally by 2 timers/counters, the serial port or 4 edge sensitive lines, but no external connection to IRQ, just an external NMI. The instruction set is the same as the Rockwell NMOS enhanced by four bit manipulation instructions (SMB, RMB, BBS, BBR).

This datasheet mentions also the R6500/13 which is the same as the R6511AQ with a 256Byte mask ROM.

- Ralf

[kakemoms]

Cray Ze wrote:

I don't think anyone has suggested to add the 600 Mitsubishi micro controllers (the thread is about cores isn't it?), which are not all mask-programmable by the way, some are external ROM, some are Flash. Many also have a full external address and data bus. There are probably five or six cores spread across the range, most of which can run 6502 code directly.

I hadn't finished with my post near the top of this thread yet, there's a lot of data and not always a lot of time, but if there's not interest, well, that's easier for me.

My point was that adding 600 flavors of the Mitsubishi 6502-core into the list is just too much work (for me). Just tracking down the number of 6502, 65C02 or 65C816 opcodes takes a lot of time. Then one also want to specify which other opcodes there are so as not to add something that is completely unrelated to 6502. And that is even more work. If you want to do it, then fine, I can add it.

If its not programmable, then you will be stuck with whatever ROM code the original programmer wanted to stick inside it. If you can run it with external memory, then great! It should be on the list. On the other end, if you CAN order it with a new mask ROM (e.g. its still in active development/support), then maybe it ought to be on the list. And PROM containing IC's should certainly be on the list.

As a limiter for the (expanding) list, I wanted to limit it to IC's that:

1) Runs the 6502,65C02,65SC02 or 65C816 instruction sets or a subset of these (like a smaller core).
2) Is not a 6800, Z80 or 8088-like processor or has instructions that make it so.
3) Can be programmed with internal or external memory without having to order 100K pcs.

For cores with a few extra instructions, I think that is fine. If its more than xx% of the total instruction set, then it starts looking less like 65xxx and more like something else. And from my perspective it becomes less interesting.

[Cray Ze]

kakemoms wrote:

My point was that adding 600 flavors of the Mitsubishi 6502-core into the list is just too much work (for me). Just tracking down the number of 6502, 65C02 or 65C816 opcodes takes a lot of time. Then one also want to specify which other opcodes there are so as not to add something that is completely unrelated to 6502. And that is even more work. If you want to do it, then fine, I can add it.

As previously mentioned, there are far fewer Mitsubishi 6502-cores than there are microcontrollers based on those cores, probably around six flavours. I'm not suggesting you do all the work, it would be better if you wait until I edit the Mitsubishi post and pick your data from there. As for listing all the extra opcodes, that doesn't fit with your list format so leave them off for now, I'll be making a dedicated thread to cover the Mitsubishi parts and their opcodes in detail with comparison tables etc. This can then be linked to from your list in order to keep it neat and tidy.

kakemoms wrote:

If its not programmable, then you will be stuck with whatever ROM code the original programmer wanted to stick inside it. If you can run it with external memory, then great! It should be on the list. On the other end, if you CAN order it with a new mask ROM (e.g. its still in active development/support), then maybe it ought to be on the list. And PROM containing IC's should certainly be on the list.

I plan to list all the user programmable Mitsubishi parts in the dedicated Mitsubishi thread, along with the current, still in production 38000/740 Series from Renesas.

kakemoms wrote:

As a limiter for the (expanding) list, I wanted to limit it to IC's that:

1) Runs the 6502,65C02,65SC02 or 65C816 instruction sets or a subset of these (like a smaller core).
2) Is not a 6800, Z80 or 8088-like processor or has instructions that make it so.
3) Can be programmed with internal or external memory without having to order 100K pcs.

Technically, the 6502,65C02,65SC02 or 65C816 instruction sets run respectively on the 6502,65C02,65SC02 or 65C816 cores, and that's the list in total, four simple entries, but this limitation goes against both the thread title and some of the parts already on your list, California Micro Devices for instance with it's 178 opcodes, or the R65C02 with 210 opcodes.

I didn't mention all the microcontrollers from Micronas (formerly ITT) that licence W65C02 and W65C816 cores because they are no different from the WDC cores, which are already listed.

kakemoms wrote:

For cores with a few extra instructions, I think that is fine. If its more than xx% of the total instruction set, then it starts looking less like 65xxx and more like something else. And from my perspective it becomes less interesting.

I think different people will have different ideas of what constitutes a 6502, for me, it's less to do with opcode count, and more to do with architectural lineage and compatibility with the original instruction set (the Mitsubishi 7900 family stretches me to the limit on this as they broke opcode compatibility, though retained the instruction set/mnemonics).

Take the Intel 8086 as an example with it's x86 instruction set, would you try to tell me that for instance, that AMD's Ryzen Threadripper isn't x86.

What is xx%? Removing the qualifier from a statement, by it's very nature, disqualifies it. Surely you have an idea of what xx% is, it would be interesting to know your opinion on how many opcodes it too many, irrespective of who may or may not agree.

[GARTHWILSON]

Make sure you also get the GTE (later CMD) G65SC150 and '151 microcontrollers.

[Cray Ze]

GARTHWILSON wrote:

Make sure you also get the GTE (later CMD) G65SC150 and '151 microcontrollers.

Would it be best to list the core just once, and have the microcontrollers using it, listed below.

G65SC00 .......c.o.r.e...d.e.t.a.i.l.s...e.t.c......................
G65SC150, G65SC151

The currently listed CMD devices could be condensed in this manor as well.

Likewise, Novatek, Megawin, and Jtech each have many 6502 based microcontrollers, though all appear to use the original NMOS instruction set on their 6502 cores.

[GARTHWILSON]

I figured they were worth mentioning since the head post separately lists the 6502, '03, '04, '05, '06, and '07 which used the same die but put in different packages (some 40-pin, some 28-pin) and bonded them differently. The 'C150 and '151 are microcontrollers though, not just bare processors. I had to deal with this in the "Availability" part of my NMOS-versus-CMOS article, at http://wilsonminesco.com/NMOS-CMOSdif/#av, and decided there were too many variations to be worth the work of rounding them all up and listing them all.

I tried to get some 151's from CMD years ago, but couldn't, because the sales woman on the phone couldn't get it through her head that I wanted 151's, not 51's. I kept saying it over and over, and she kept repeating things like "Got it: ten [or however many it was] G65SC51's," and I would repeat, "No, not fifty-ones, but one-five-one!" I thought she finally understood, but then in the mail here come fifty-ones. I gave up.

[Druzyek]

This is a really cool list! Thanks for working on it.

kakemoms wrote:

BigEd wrote:

Fantastic list! The much more recent SunPlus (?) and I think something from STM (SGS-Thomson) are subsetted 6502 with a register missing.

If you have any more information on SGS-Thomson, feel free to add it!

Apparently the ST7 is derived from the 6805, not the 6502: http://www.cpushack.com/2013/05/21/mcu- ... 05-reborn/

Quote:

SGS Thomson "ST7 series" according to the linked page. I looked at the ST7 instruction set and architecture, there are some similarities, but I'm not sure I'd call it a 6502 variant, a cousin perhaps.
http://members.casema.nl/hhaydn/howel/p ... 02_CPU.htm

From the page you linked: "These are based either on the 6800 or 6502 . . . The ST7 stores 16-bit pointers with the most significant byte at the lowest address, the opposite to normal 6502 convention." 680x series chips store the the most significant byte first

Quote:

we can´t add 600 different mask-programmable Mitsubishi controllers

Agreed but I definitely think we should point out the most important aspects:
-Multiply and divide instructions as Cray Ze mentioned
-X index addressing mode lets you use any zero page address pointed to by X as an accumulator. This is enabled with the new T flag.
-The M50734 for example seems to divide its clock by 4. You can see two dividers on page 29 of the datasheet: http://pdf1.alldatasheet.com/datasheet- ... SP-10.html. Also, minimum instruction time at 8MHz is listed as 1us on the first page of the datasheet, which I think comes from 2 cycle minimum instructions and dividing by 4. Good to know you are only getting 2MHz performance at 8MHz before you put one of these in your project!
-The M50734 is ROMless and the address bus is multiplexed (page 1).
-Another Mitsubishi example, the M50747, has internal memory but can be set to run from external memory by software or by changing a pin (see page 20): http://pdf.datasheetcatalog.com/datashe ... 916_1.pdf.
-Page 22 shows the same divide by 4 arrangement as the M50734 but the address bus is not multiplexed (page 20).
-The M50740 seems to be the ancestor of the Renesas M38000 line. None of the 3800 datasheets I found on their site showed an external bus option. Flash models are limited to 100 writes ( ) and some of the older ones require 12v to program.
-The "3800 Group User's Manual" (https://www.renesas.com/en-us/doc/produ ... 3800um.pdf) lists processor mode on page 100, so some of the line must have been able to work with external data at one point.
-Page 45 makes it look like that series divides the clock by two instead of four.

How about adding the HuC6280 to the list? https://en.wikipedia.org/wiki/Hudson_Soft_HuC6280

[BigEd]

Certainly worth adding the HuC6280!

[kakemoms]

HuC and a couple of others added: Ricoh 5A22, Nintendo SA1, UM6502(A/B/C), UM6507, UM6512.

Some info also adjusted according to posts.

[kakemoms]

I added Synertek which for some strange reason was not on the list..

[GaBuZoMeu]

I think the W65C265S is missing.


Merry Christmas to you all!

Arne