Z80 vs. 6502

[JamesD]

After giving the RAM/ROM speed issue some more thought. There were probably higher speed RAMs available but they would cost more. At what point in time they were high enough speed and when they became affordable I don't know. I seem to remember DRAMs as fast as 120ns being available in the early to mid 80s but I could be wrong.

 

My Apple II+ has 4116-20NL DRAMs. 200ns and NMOS Logic? That's probably the same RAM as in the Apple II so it had been around since at least 1975.

I *think* the 64K RAM upgrade for my CoCo was 150ns DRAMs and that would have been 1983. I think we paid around $50 for the chips so you know they had been around for a while.

<edit>

The IIgs motherboard I have here uses 120ns memory chips.

Edited December 17, 2012 by JamesD

[carmel_andrews]

the nice thing about both processors is that they initially kept mototrola and intel out of the home pc/computer market

 

At least those that were interested in extending the lifespan of both processors like WDC and Zilog themseves saw the potential of upscale versions of said processors, like the z80000 and the 65816

[Rybags]

I don't particularly see what good there was in keeping Motorola out.

 

The 68000 was probably the single best ever CPU in terms of comparisons to the competition of it's time.

 

The so-called "future" versions of Z80 and 6502 were pieces of crap in comparison to the 68000, let alone the later variants like 68020/30/40.

[Dr Do]

You know, listening in to this conversation has made me wish I hadn't thrown away my A-Level Computer Studies notes when I took up a different subject at uni. One of the things we learned about was CPU operations but it was so long ago that I've forgotten a lot of it. It was interesting stuff though!

[JamesD]

the nice thing about both processors is that they initially kept mototrola and intel out of the home pc/computer market

 

At least those that were interested in extending the lifespan of both processors like WDC and Zilog themseves saw the potential of upscale versions of said processors, like the z80000 and the 65816

I really don't understand why you would want to keep Motorola and intel out of the market.

 

I don't particularly see what good there was in keeping Motorola out.

 

The 68000 was probably the single best ever CPU in terms of comparisons to the competition of it's time.

 

The so-called "future" versions of Z80 and 6502 were pieces of crap in comparison to the 68000, let alone the later variants like 68020/30/40.

Rybags is right. The the Z80 and 6502 were crap compared to the 68000.

Even the 6809 looked like crap compared to the 68000 and the 65816 isn't even as nice as the 6809.

 

The 68000 programming model still holds up well to this day as the basis for the Coldfire processor line which are very popular in embedded systems.

While the 68000 register set isn't completely orthogonal, it's not bad.

The CPU predates RISC so it leaned more towards supporting compilers rather than simplifying things for pipelining. However, superscaler architecture wasn't around yet when RISC was invented and compiler technology has advanced greatly so much of the initial performance advantage of RISC was short lived. By dropping some of the more complex 68000 instructions Freescale was able to take the 68000 to RISC like performance and with greater code density. Too bad Motorola had been too set in their ways to move in that direction sooner.

 

The Z800 and Z80,000 were never released and pretty much have to be relegated to the land of speculation like the 6509. The only reason I like to talk about the 6509 is because it would have made a big difference on existing machines and it would have been released in the mid to late '70s. Much of the 65816's design borrows from the features listed for the 6509. The Z80,000 wasn't even prototyped until 1983 if I remember right so even if it had been released you probably wouldn't have seen a machine based on it until '85 anyway.

 

On the other hand, the Z8000 was certainly released. It is not object code compatible with the Z80... but that is probably a good thing.

It has 16 - 16 bit general purpose registers which makes compiler support easier and the programming model nicer than the Z80.

But the Z8000 is still just a 16 bit CPU with an MMU. 16 bits = 64K + MMU = 8MB. Zilog has CPUs with and without the MMU.

But the Z8000 is no match for the 68000 with it's flat memory model and full object code compatible 32 bit CPUs in the family.

The closest CPU I can think of to the Z8000 was the TI9900, but the Z8000 has internal instead of external registers, and a better instruction set for the most part.

A machine based on the Z8000 might have been a tough competitor in '79 when the CPU was introduced but I don't think it was priced for the personal computer market.

If you want to start a what if topic, the Z8000 would certainly be a good subject for discussion.

[Arkhan]

I started with 6502 (C64 and then PC Engine/Turbo Grafx-16 and some Apple II).

 

I then moved to Z80 with the MSX, and found that z80 is alot easier to work with, at least to me. I like the fancy-ass register setup more than the zero-page setup, and also find that there are alot of other neat instructions and things you can do with the z80 that the 6502 cannot.

 

One of the things I like is that loads on the Z80 don't affect the flags. That was probably the biggest "!!!!!" moment when I switched to z80.

 

The zero-page setup of 6502 gives me a headache, and so does the double indirect indirection to indirecty indirectness addressing mode.. The only 6502 I really enjoyed was the PC Engine because it's got some extra facilities in it that make it less of a headache.

 

In the end though, neither is really absolutely better than the other. I've read various books and articles breaking down various points explaining how the 6502 could perform better because of how it pipelines instruction execution, and how the z80 could do better because it has real 16 bit registers and whatever else, but really, if you use either one as intended (or even close to being as intended), you're going to get about the same kind of performance, so the real thing that will matter is the video/sound chips that are involved, and even that will boil down to preference.

 

If one were radically more powerful than the other, you wouldn't have double dragon on SMS and NES both play about the same with the only difference being the graphics and sound.

 

Another example: People rant and rave about shoot em ups being great on the C64 but I think they all suck for the most part (Especially Lions of the Universe. aoifjsdfoijasdfoasf I hate that game). I can't think of any I really liked alot that weren't somewhere else, except for Parallax, if that even counts. It might be because I played all these games on NES/PCE/Sega Genesis/Amiga first, so the C64 ones are a bit bland in comparison... Side Arms sticks out there as one of the blandest.

 

I also prefer music in the PCE games a bit more than everything else because I prefer that specific sound. You'll find other people who prefer SID tunes the same way. It doesn't really matter though. All them old machines are like boobs. All boobs are good. Just pick the ones you like fondling the most.

 

I'm glad this thread has gone 5 pages so far without turning into a flamewar disaster. I was expecting chaos by page 3.

 

Usually these discussions turn into people posting C64 demos while making fun of spectrum ones, and vice versa. With neither side realizing how stupid the argument has become.

[JamesD]

I got to read a lot of the BYTE magazine scans over the holiday. If anyone wants to know when chips came out or how much they cost it's a pretty good resource.

The Z80 definitely came out after Woz designed the Apple 1 and when it did, sample chips were $200 each from what one article said.

Everything was pretty much 8080 or 6800 in the Apple 1 days.

[Rybags]

8080 was sort of to the Z80 as the 6800 was to the 6502 - Compucolor II used the 8080, it was one of the first consumer-level colour computers.

 

Funnily enough, the x86 is a descendant of that chip, part of the reason the instruction set and overall architecture is such a mess.

[carlsson]

That vaporware 6509 CPU mentioned early on in this thread surely was supposed to be another beast than the 6509 CPU that actually emerged from MOS/CSG? You can find it in the CBM-II series (B128, B256, B500, P500, CBM 610, CBM 710 and so on depending on whether you're in the USA or Europe). It is to most part a 6502 CPU but with bank switching abilities to address up to 1 MB IIRC. Given that the CBM-II series were released in parallel or even slightly before the C64, I believe the 6509 and 6510 (which is used in the C64 - itself being a 6502 with an 8-bit I/O port to bank switch ROM on top of RAM but often misquoted as having additional abilities) were available at the same time. It could though be noted that Commodore apparently didn't sell any 6510 or for that matter other newly produced custom chips to competitors, only the earlier chips from the MOS years.

 

Btw, for a 8-bit RISC CPU some people sometimes refer to the RCA 1802 but I figure that is mostly for laugh as it is quite slow and I think needs additional circuitry to be used anyway. When people in modern times think of RISC, they usually think of something rather fast and powerful, not something dog slow even compared to its contemporary competitors. Actually there was a thread on the Vintage Computer Forum some years ago where hobbyist veterans who actually twiddled with computers already in the early 1970's discussed the offerings on early CPU's, which ones were more notable, affordable and usable than others. The discussion however focused on the years right before the 6800, 8080, 6502 and Z80.

[Keatah]

They shoulda made a 25MHz 6502 for the Amiga. Then it mighta beena nice computer. Bwaaahahaaa! I'm outa here..

[JamesD]

That vaporware 6509 CPU mentioned early on in this thread surely was supposed to be another beast than the 6509 CPU that actually emerged from MOS/CSG? You can find it in the CBM-II series (B128, B256, B500, P500, CBM 610, CBM 710 and so on depending on whether you're in the USA or Europe). It is to most part a 6502 CPU but with bank switching abilities to address up to 1 MB IIRC. Given that the CBM-II series were released in parallel or even slightly before the C64, I believe the 6509 and 6510 (which is used in the C64 - itself being a 6502 with an 8-bit I/O port to bank switch ROM on top of RAM but often misquoted as having additional abilities) were available at the same time. It could though be noted that Commodore apparently didn't sell any 6510 or for that matter other newly produced custom chips to competitors, only the earlier chips from the MOS years.

Definitely a different 6509. It was something a 2nd source 6502 manufacturer leaked to MICRO the 6502 Journal.

When the magazine published an article about it, the company denied all knowlege of the beast and it was never mentioned again.

Funny thing... a lot of the things mentioned ended up in the 65809/65816.

Dump the 24 bit addressing, some special mode stuff and you have a pretty good idea of what the 6509 was supposed to be.

 

Btw, for a 8-bit RISC CPU some people sometimes refer to the RCA 1802 but I figure that is mostly for laugh as it is quite slow and I think needs additional circuitry to be used anyway. When people in modern times think of RISC, they usually think of something rather fast and powerful, not something dog slow even compared to its contemporary competitors. Actually there was a thread on the Vintage Computer Forum some years ago where hobbyist veterans who actually twiddled with computers already in the early 1970's discussed the offerings on early CPU's, which ones were more notable, affordable and usable than others. The discussion however focused on the years right before the 6800, 8080, 6502 and Z80.

RISC is clearly a very different beast than pretty much anything that came before it.

Many of the pieces had been around for years such as pipelining, or even load and store architecture, but I think the biggest difference was how the hardware and compilers relate to each other.

 

I think the closest thing to RISC that I've seen in a microprocessor prior to the research I mentioned would be the Z8000.

It's registers are *mostly* general purpose 16 bit (with 32 bit support) and it's sort of a load and store design.

I think some of the instructions are a bit complex and I would say it's close but not quite RISC.

I don't believe the initial models were pipelined but I think some later chips used pipelining.

If Zilog had priced the Z8002 right it would have made for a nice 1980 PC and could have lead to followup machines with more than 64K.

[JamesD]

I've been reading through old issues of MICRO looking for info on a couple topics and I ran across some info related to this thread.

 

First of all... I located the info on the 6509.

The 6509's actual name was to be the SY6516 and the manufacturer was Synertek.

 

The first mention of the SY6516 that I ran across as in issue 21 (Feb 1980) Page 11.

The 6516 might have been mentioned months earlier because the part was supposedly announced some time in early 1979.

The text is mostly a list of instructions that take fewer clock cycles on the SY6516 and as far as I could tell and it seemed to be tacked on to the end of another article.

The 6516 was to be source code compatible with the 6502 but not binary compatible so it turns out the idea of just plugging in an improved chip is pure fantasy on my and other people's part. A binary compatible chip would certainly have been possible but I think thee incompatibility was to avoid legal problems. It's also possible they had a way of encoding the instructions that made decoding easier but I guess we'll never know.

For a new machine that doesn't need to be backwards compatible, object compatibility wouldn't have been a problem.

 

Issue 22 has a teaser on page 22 about the upcoming article and asks other people to write in what they think the chip needs.

 

Issue 23 has the main article on page 36. The article claims the chip was designed by (for?) Atari but Synertek wasn't able to deliver.

The chip was named the 6509 when it was originally designed.

The article has a lot of details and the author mentions an EDN article about the chip but I am unaware of any scans of EDN back issues from that time.

The author (Randall Hyde) stated he was writing an assembler for the chip and it turns out he wrote the Lisa Assember for the Apple II... which is what I learned 6502 assembly on. He also wrote SmartBasic for the Coleco Adam and I was just looking through the disassembly of that two days ago!

 

Issue 24 has the reply by Synertek on page 7 where they basically say it doesn't exist. In some issue after that I know Mr. Hyde responds but I didn't find the actual response which is all I think I had read previously. I believe he said something about having the data sheets and I even think he had spoken with or had heard from someone from Synertek prior to writing the article but I can't be sure.

 

The issues that follow publish some reader comments on what an enhanced 6502 should have and in issue 34 there is an article about a 6502 dream machine on page 69.

 

Issue 37, the dream of an improved 6502 is dead and the magazine starts supporting the 6809 as well as the 6502.

 

 

 

As for the other info I ran across...

Multiprocessing on an 8 bit.

 

It turns out that the Z80 cards take over the Apple II so no multiprocessing can take place with them.

I found some info on a 68008 card but it didn't say whether it took over or not.

However, shortly after MICRO starts supporting the 6809 an advertisement appears for "THE MILL" which is a 6809 board.

You can find the ad in MICRO 41 page 75 but it may be in earlier issues.

The ad specifically mentions the ability to multiprocess. The 6809 runs at full speed and the 6502 runs at 20% speed when the 6809 is enabled.

 

 

 

BTW, there was also a card that had a CPU that ran PCODE directly and could speed up execution of Apple Pascal programs.

It was advertised around the same time as THE MILL.

[+nanochess]

There is a rumour that the Z800 was so powerful that the USA military took it for its missiles.

 

Anyway Zilog finally released it in 1989 as the Z280, it provided protected memory, bank switching and a lot of other things. It could have been very good for a 16-bits UNIX like the one in PDP-11 or even Minix, but unfortunately it was too late. Finally the Z280 was discontinued my middle 1990's and completely erasen from Zilog history (now it's very hard to find any info)

 

I've saw also a Z380 that had 32 bits registers, it doesn't had success and also was wiped out.

[JamesD]

There is a rumour that the Z800 was so powerful that the USA military took it for its missiles.

 

Anyway Zilog finally released it in 1989 as the Z280, it provided protected memory, bank switching and a lot of other things. It could have been very good for a 16-bits UNIX like the one in PDP-11 or even Minix, but unfortunately it was too late. Finally the Z280 was discontinued my middle 1990's and completely erasen from Zilog history (now it's very hard to find any info)

 

I've saw also a Z380 that had 32 bits registers, it doesn't had success and also was wiped out.

I don't have any direct knowledge of the Z800 being used or not used in military applications, but I do think for missiles a DSP or custom IC would be more appropriate than the Z800.

Several websites have stated the Z8000 was used for military applications and some form of it is supposedly still in the Zilog catalog. That may be where the rumor comes from about the Z800.

 

I have my doubts about the Z800 and UNIX but some form of enhanced CP/M or MP/M would make a lot of sense.

The whole reason for backwards compatibility is to run old apps or allow easy migration to the new chip at least at the source code level.

CP/M and MP/M encourage migration of software but UNIX requires a new code base. Unix is also commonly associated with C and I'm not sure the Z800 register model was a significant improvement in that regard.

 

Frankly, the Z8000 was much better for Unix than any Z80 derivative and the Commodore 900 probably would have shaken up the Unix world if it had been released.

The Z80000 would have made for a heck of a followup machine and intel and Motorola would have been well behind Zilog in features at that point.

Funny thing... if Commodore had released the 900 it might have cured financial problems that CBM and Zilog faced but some manager cancelled a project that was supposedly ready to ship because Commodore had acquired the Amiga. The products didn't compete with each other so I really don't understand the decision.

[Arkhan]

I have an MSX Turbo R, with an R800 in it.

 

 

That thing's stupid-fast.

 

I like it.

[carlsson]

Mr. Hyde [..] said something about having the data sheets

Perhaps those were early drafts that somehow were leaked to him or someone else in the industry? In particular if Atari had custom ordered the CPU, they should have something to say about it as well, or that they silently bit the bullet and ordered 6502 chips instead.

 

Anyway, February 1980 is a bit late if the CPU was meant to be used in Atari's computers from the start. Early 1979 obviously fits better with the parallel development of the ANTIC and CTIA chipset.

 

I found an excerpt from the book "Atari Inc - Business is fun" by Marty Goldberg and Curt Vendel. I don't know how you in the Atari community relate to this book, but Curt to me seems like a reputable person? In the chapter about Atari computers, it starts in February 1977 with the "New Machines" project. Some time in the late spring, Atari supposedly pretended to produce an Apple II clone but in reality worked on their own design. In July 1977, Atari evaluated MOS 6510 (the CPU used in the C64 five years later!) as well as the 6560/61 VIC-I chip (used in the VIC-20 in 1980/81). It says the VIC chip would produce a text display of 36 columns x 14 lines which to me sounds like a rather different chip than the one I've been programming for the past 15 years. The text even says Atari briefly considered a TMS9900/9940 but it was too expensive. In September 1977, Jay Miner receives a report on the Colleen that it might use a 6502 or a 6509, something not yet decided by December the same year. In February 1978, the book says that Atari for two days (!) considered designing their own 6509. By the end of March they've even considered a 6809, but eventually sourced 6502 from Synertek.

 

The earliest datasheet I've found for the 6510 is November 1982, but since the C64 was released months earlier there ought to exist earlier datasheets too. Perhaps not as early as 1977. The same about the VIC-I chip, which I've read was designed by Al Charpentier in 1977 but Commodore never found a market for. If the above Atari book is entirely correct on every detail, it would suggest that someone at Atari went to MOS laboratories to investigate a brand new, perhaps not even finished video chip while at the same time working on their own chipset to roughly have the same functionality. Since Commodore bought MOS in late 1976, it should have spurred Jack Tramiel if he had been made aware about a major video game manufacturer visited his company to look for parts for a computer, and why not Commodore themselves tried to produce a colour PET much earlier than the TOI project.

[JamesD]

I have an MSX Turbo R, with an R800 in it.

 

 

That thing's stupid-fast.

Web pages say the R800 is based on the Z800 but I don't remember it having all the upgrades the Z800 supposedly has... but then I haven't programmed either one.

 

The R800 is a fully pipelined Z80 with a 16 bit ALU and added instructions.

It's definitely fast for a Z80 but some of the Turbo Rs speed comes from a fast clock speed.

I think the R800 is around 2.5 times as fast as the Z80 at the same MHz.

The 64180 was around 20% faster than the Z80 on the same code just due to the prefetch and 16 bit ALU so you can clearly see what the pipeline does for the R800.

That *probably* makes the R800 faster than the 6502 at the same MHz.

I *think* the 65816 and R800 are almost a tossup speed wise thanks to the addition of 16 bit support in the 65816.

The 6809 should be faster at the same MHz than the R800 but even if it's not, the 6309 is about 20% faster than the 6809 on the same code once you enable native mode so a 6809ish machine would still be faster.

Disclaimer: This is all based on claims that the Turbo R is 5 times faster than the MSX2, the R800 is clocked at twice the speed of the MSX2 Z80, the Z80 needs 2.2 times the clock speed of a 6502 to run the same speed and the 65816 and 6809 are faster than a 6502 at the same clock speed. Which is another way of saying I'm guessing.

 

Some people have a 14MHz or faster Apple IIgs and the CoCo 3 in an FPGA at 25 MHz runs BASIC almost as fast as if it were machine language. If only we could have had machines like that in the early '80s!

 

The R800 doesn't do a lot to reduce code size so it probably only beats the 6502.

I like it.

So do I, it's why I bought the Turbo R.

I should take more time to play with it.

[JamesD]

Perhaps those were early drafts that somehow were leaked to him or someone else in the industry? In particular if Atari had custom ordered the CPU, they should have something to say about it as well, or that they silently bit the bullet and ordered 6502 chips instead.

<snip>

In September 1977, Jay Miner receives a report on the Colleen that it might use a 6502 or a 6509, something not yet decided by December the same year. In February 1978, the book says that Atari for two days (!) considered designing their own 6509. By the end of March they've even considered a 6809, but eventually sourced 6502 from Synertek.

Interesting!

Well, I think some amount of engineering a 6509 must have taken place at Synertek before the Atari 8 bit was completed or it wouldn't have been mentioned in the book.

The project would have to at least be through a significant part of the design phase in order to know which instructions were faster on the 6509 than on the 6502 in order for them to be published in MICRO.

It's certainly well beyond the wish list phase that the letter from Synertek indicated.

 

I don't think 6502 programmers really appreciated just how much better the 6809 was than the 6502.

In the article about the 6516, the author seems to think the 6516 is better than the 6809. But when 6809 articles started to appear in MICRO, there were a few comments that seemed to indicate people were starting to see the light.

If Atari didn't really get the advantages of the 6809, they would obviously choose the 6502 on the basis of cost alone.

[carlsson]

I hoped to find some early reference to how much a 6809 would cost, but none found yet. The earliest reference in BYTE magazine is from January 1978 which has an article by Carl Helmers on page 168. There are some more articles about Motorola's new hybrid 8/16-bit processor later that year, but I suppose pricing mostly was based on individual quotes. Assuming the 6502 still could be had for $25 or less (now second sourced from Synertek, did they have the same prices as MOS did?), it might very well be so that the Atari dev team quite didn't see what processing benefits a more advanced processor would give their still unlaunched computer.

 

More importantly, was there an opportunity for marketing to explain an extra $30-$50 or more because this computer has a vastly more powerful brain than the Apple II, Commodore PET, TRS-80 etc? In particular as I tend to remember that the very first 400 and 800 had rather little RAM (upgraded later), what would the customers think if the manufacturer rather spent extra on the latest CPU technology than workspace memory to run the programs?

Edited January 16, 2013 by carlsson

[carlsson]

As a somewhat off-topic addendum, I browsed through Creative Computing Magazine from 1978. Archive.org only has four out of six issues from that year, but neither of those ever mention the 6809 a single time, at least not per the auto-OCR:ed text. I don't know how Creative Computing and BYTE differed in content and focus groups, if new hardware that not yet had made it to practical computers would be irrelevant to CC readers, and they'd rather focus on articles what you can do with the computer you already got.

 

I suppose the MICRO magazine that JamesD went through above was Micro 6502 Journal? Issue 37 would equal June 1981, so three years after the 6809 was launched. For that matter, 68 Micro Magazine was launched in the beginning of 1979 but I haven't studied if the two were related. COMPUTE! also was launched in 1979 as an extention of the PET Gazette, which might discuss the 6809 only in terms of the SuperPET 9000 in 1981.

 

Additionally I looked for references to a 6509 in BYTE 1977-1978. I don't know whether they came across and published rumours. Unfortunately the only references I found were false positives relating to the zip code 06509 in New Haven, CT. If I have the time, I might look into 1979-1981 issues and search both for 6509 and 6516.

[carmel_andrews]

Re: original VIC chip (not vicII), just after the Pet was launched, commodore got MOS to try flogging (selling) the VIC chip to competing hardware companies (see 'home computer wars' for referrence), apparently they weren't that successful at it, so commodore decided to get in on the act and the earliest vic chip based design were meant for Pet based systems

 

The Vic 20 was merely an afterthought, when Peddle kept dragging his feet over the Pet based Vic chip systems, Peddle wanted a system to compete with Atari and Apple and tramiel wanted a low end system that cost less then the Atari, Tramiel won the argument and the pet was out of the picture so far as the vic chip was concerned

 

Another interesting thing is, according to Joe Decuir, the CTIA was already finished (and ready for use) by the time the VCS was released

[JamesD]

I hoped to find some early reference to how much a 6809 would cost, but none found yet. The earliest reference in BYTE magazine is from January 1978 which has an article by Carl Helmers on page 168. There are some more articles about Motorola's new hybrid 8/16-bit processor later that year, but I suppose pricing mostly was based on individual quotes. Assuming the 6502 still could be had for $25 or less (now second sourced from Synertek, did they have the same prices as MOS did?), it might very well be so that the Atari dev team quite didn't see what processing benefits a more advanced processor would give their still unlaunched computer.

 

More importantly, was there an opportunity for marketing to explain an extra $30-$50 or more because this computer has a vastly more powerful brain than the Apple II, Commodore PET, TRS-80 etc? In particular as I tend to remember that the very first 400 and 800 had rather little RAM (upgraded later), what would the customers think if the manufacturer rather spent extra on the latest CPU technology than workspace memory to run the programs?

Well, I think they overlooked the fact that the 6809 requires about 1/3 the code of the 6502. That would have benefited Atari and consumers.

People could squeeze larger programs in the same RAM space, larger games in ROM carts, and more in the BASIC cart.

A BASIC with more commands requires less RAM for BASIC programs. Just adding ELSE to the IF THEN clause cuts the number of lines of BASIC code, and tends to make code more readable and run faster.

6809 code also takes less time to write. Face it, no messing with page zero just to deal with 16 bits, fewer lines of code do deal with the stack, auto increment/decrement, not having to spend a dozen instructions to calculate an address that the 6809 can calculate in 1 instruction (this was an example in MICRO), etc...

I think the 6502 programmers didn't really see the advantages and probably just focused on how they could do everything on the 6502.

 

And then maybe they didn't negotiate with Motorola over price.

How expensive could it have been if Tandy was able to base a budget priced PC on the chip?

[Csonicgo]

I played a Rogers Z80 organ for about 5 years. I assume there are more instruments out there using the Z80 that I don't know about. I know Yamaha used the Motorola 6800 in their DX series, and the 6502 in cheaper FM synth keyboards. So for single-purpose applications, either one works.

Edited January 17, 2013 by Csonicgo

[carlsson]

the 6809 requires about 1/3 the code of the 6502

Wow! I've never studied the 6809 instruction set in detail. I knew it is more powerful, but not that the difference on average would be that big. How many of the original Atari 8-bit releases were written by VCS developers who would've had an increased learning curve to at least somewhat take advantage of the different mindset? I realize 2600 games don't port themselves to the 8-bit computer line anyway, so perhaps that was a non-issue.

 

The budget priced Tandy, are you referring to the "Green Thumb" Videotex terminal started in 1977 and released in late 1979?

 

Some crude comparisons:

 

TRS-80 CoCo 1: 0.89 MHz 6809E, 4K RAM. Announced July 1980, launch price $399

Atari 400: 1.79 MHz 6502, 8K RAM. Announced December 1978, launch price $499 (or if it was $549 or $599)

Commodore VIC-20: 1.00 MHz 6502, 5K RAM. Announced 1980, launch price $299

 

I find it quite likely that Motorola might have dropped their prices quite a bit between the launch in mid 1978 and two years later when the CoCo was announced.

 

Furthermore, I realized that Google has indexed most issues of InfoWorld. The earliest reference to a 6809 is an advertisment from May 1979 where SWTPC advertised a 6809 computer. The processor card alone was $195 and a complete computer $1500. In October the same year, Percom Data Company launched SBC/9, a card for their series of SS-50 bus microcomputers that also could be used as a single board computer. The SBC/9 with 1K RAM, RS232 and two EPROM sockets (firmware not included) was $199.95 on its own. Frankly it doesn't say much. Compas Microsystems in June 1979 launched their CSB 1 which had a 6502, a 6522, two 6520, 2K RAM and four EPROM sockets. Evaluation price for that system was $395, regular price $595. Compared with the SBC/9, suddenly a 6502 system seems rather expensive but those were two different SBC designs from two different companies and with different capacity.

 

Now for the $1000 question: what does InfoWorld have to tell about the 6509 and/or 6516? Well, I found exactly two mentions:

 

6509 or the 6516

 

Rumors about the impending introduction of an upgraded 6502, to be called either the "6509" or the "6516", and which would have 16-bit registers and expanded addressing, are decidedly premature, according to all three major suppliers of the 6500 series.

 

Commodore, the owner of MOS Technology (the original supplier of 6502's) said that their present production is concentrating on ROM and RAM chips, and that CPU development lies in the future since the design resources are not available.

 

Synertek, who had made preliminary design studies for the enhanced chip, stated that the development had been indefintely postponed while they concentrated on developing peripheral controller chips and second-sourcing the Z-8.

 

Rockwell, the third supplier, states that although they are developing a line of enhanced 6500 series chips with RAM and ROM on-chip, the joint development plans with Synertek on the 6516 have been postponed. All expressed willingness to second-source the 6516, if someone else would develop it.

 

Look for two new Apple machines to be introduced shortly. One is the Apple IV, a low-end machine like the Apple II, but the speculation is that it meets the new RFI specs. The other is the Apple V, aimed at the scientific/professional user. The Apple V is rumored to contain the much-rumored 6516 - the 16-bit 6502.

 

Actually this last rumour isn't so far off if you replace Apple V with Macintosh and 6516 with 6809, which the earliest prototype boards supposedly had before Apple moved to 68000.

[JamesD]

Wow! I've never studied the 6809 instruction set in detail. I knew it is more powerful, but not that the difference on average would be that big. How many of the original Atari 8-bit releases were written by VCS developers who would've had an increased learning curve to at least somewhat take advantage of the different mindset? I realize 2600 games don't port themselves to the 8-bit computer line anyway, so perhaps that was a non-issue.

Actually, I think that is a key issue. I think 6502 devs liked what they had and didn't really truly give the 6809 an honest look.

 

The budget priced Tandy, are you referring to the "Green Thumb" Videotex terminal started in 1977 and released in late 1979?

Actually, that terminal WAS a stripped down CoCo if I remember right. Same basic hardware.

 

I find it quite likely that Motorola might have dropped their prices quite a bit between the launch in mid 1978 and two years later when the CoCo was announced.

That is likely but for a huge design win they might have dropped the price a bit sooner. I doubt prices were over $75 at introduction even

in small quantities. I'd guess it commanded a $25 premium over the cost of a 6800.

 

Furthermore, I realized that Google has indexed most issues of InfoWorld. The earliest reference to a 6809 is an advertisment from May 1979 where SWTPC advertised a 6809 computer. The processor card alone was $195 and a complete computer $1500. In October the same year, Percom Data Company launched SBC/9, a card for their series of SS-50 bus microcomputers that also could be used as a single board computer. The SBC/9 with 1K RAM, RS232 and two EPROM sockets (firmware not included) was $199.95 on its own. Frankly it doesn't say much. Compas Microsystems in June 1979 launched their CSB 1 which had a 6502, a 6522, two 6520, 2K RAM and four EPROM sockets. Evaluation price for that system was $395, regular price $595. Compared with the SBC/9, suddenly a 6502 system seems rather expensive but those were two different SBC designs from two different companies and with different capacity.

Prices for SS-50 or similar systems continued to be much higher than personal computers well after that and vary a lot from one manufacturer to the next so it's tough to use that as a basis for comparison.

 

BTW, there was actually a portable 6502 machine announced early on that was similar to a SOL but in a portable case. I don't know if the company ever actually produced any machines besides the prototypes though.

A lot of the popular 6502 systems in the early days (and least from the look of MICRO) seemed to be single board machines like the KIMs, AIMs, and superboard .

The good old 4K of RAM is huge days.