Cerberus 2100 (and 2080)

[paulb]

From this thread:

OLIMEX is now selling the bi-processor Z80 + W65C02 64Ko = CERBERUS2100 at 220€:
For those interested, the Hardware Manual describes well the board.

I was looking at this today while digging around to investigate the Agon Light documentation as featured in another thread. (Yes, I know that is a different project.)

It isn't a bad overview of the hardware, I suppose, although there is an element of "white paper" about it. I was particularly amused by the very first sentence in the introduction:

CERBERUS 2100™ is an innovative, fully-functional computer, not a toy

The Agon guy really has a hang-up about toys. I am sure it is a proficiently designed product, but I find several of the choices bemusing, like the insistence on a combination of 6502 and Z80 processors, particularly when emphasising the multi-processor capabilities. I know that there is a strong appeal to authenticity here, but I think that many people back in the day sought to move up to the 68000 and other processors when putting several of them in a machine.

Using the 68000 (or, really, the 68010 onwards to get functional page fault handling) would provide broader opportunities for that educational aspect that people like to emphasise with these efforts, as I noted with the Commander X16, although the creator of that doesn't really make any strong representations about the educational value of his system. The Cerberus' graphical capabilities seem pretty underwhelming: 320x240 with a character-based display organisation, and the screenshot features the horrible Spectrum character set.

Given that a lot of these projects have been instigated to some extent by the existence of the Colour Maximite single-board computer, which still delivers a reasonable experience using off-the-shelf parts, I think that they really have to demonstrate comparable levels of utility and usability. At the low end, I suppose that the Neo6502 and Agon Light have both presented themselves as a kind of Maximite successor. And if charging larger sums of money, there has to be additional utility and, if presented in an educational context, sufficient accessibility for educators and students.

Otherwise, for 219 EUR, this comes across as a bit of a vanity project, subject to the kind of disparaging remarks about toys familiar to its creator. Not that I begrudge anyone their vanity projects, it should be noted: I have plenty of them myself.

Edit: I've updated the thread title to reflect that the commentary is largely about the new Cerberus 2100, which is a refinement of the Cerberus 2080, of course.

[komadori]

My first thought on the Cerebrus is that had _I_ already designed a system that was enjoying the moderate popularity of the Agon-series, I would focus on growing the software and documentation ecosystem of the Agon rather than on producing an apparently incompatible (?) successor. I suppose the danger of enjoying designing hardware platforms is that you're prone to doing it repeatedly. Not that there's anything wrong with doing that for one's own sake, but if you want an audience to enjoy and build on what you've created that it seems counterproductive.

As it happens, I starting writing up some thoughts on my blog yesterday about the ongoing value of retro-computing platforms compared to modern ones and the place of "neo-retro-computers" like the Agon. I made the point that, from a certain kind of software perspective, it's positively an advantage that Acorn isn't around to make new and incompatible variants of the BBC Micro. There's a timeless quality to it and the Advanced User Guide is never going to be out of date.

I'm flush with ideas so I'm definitely in danger of doing an XKCD #927 and coming up with an entrant to the space myself. In that case though, I would definitely make writing a tome equivalent to the Advanced User Guide the cornerstone of such a project. Also, and promise not to create an incompatible successor.

I do find the design decisions of the Commander X16 a bit bizarre, but the Agon guy's rant doesn't flatter him. He'd be better off focusing on the positives of his own product!

[gordonDrogon]

Hesitant to post about this, but here goes...

The Cerbreus was designed/built before the Agon. I had an early model with the task (should have kept my moth shut) of porting BBC Basic to it.

I did this - starting with my RubyOS - which is an Acorn MOS compatible OS for the 65C02 which I wrote for my own 65C02 (and now '816) SBC projcts. I added framebuffer support to it and crude graphics (the initial release used a reprogrammable font for big pixels) BBC Basic 4 worked perfectly well without any changes to the ROM image.

(I gather the latest release has slightly better graphics and colour - this was mono, 40 columns, 8x8 font)

That was that apart from a few things - I found a hardware bug in the process (oddly enough the same issue I had on my own SBC) and Bernardo wasn't happy - I'd have to dig through my emails but I had a workaround which he latterly fixed in the CPLDs but he wasn't happy and blamed the crappy 6502 - probably why he stuck with the Z80 for the Agon...

Anyway as it was nearing completion I said I'd not be releasing my code under any open source license but a source available one and that was the last thing I heard from him. Not a peep more. I was a bit miffed especially as I'd put the work into it even bought an expensive programmer which I was unable to use under Linux.

The whole project was a bit weird anyway - why bother? Was there really any educational value in this dual but mutually exclusive CPU setup? Not IMO but there you are. There was also a couple others working on it too - in addition to the Z80 there is an ATmega which does the host and filing system side of things (not a million miles from my own SBCs really) and no-one could agree on some common code for that - the 6502 and Z80 sides required different ways to boot, etc.

Subsequently reading his public comments on the X16 reaffirmed my decision to not pursue it further after he stopped replying to my emails. I don't feel his comments were at all necessary, so decided in light of that to not even think about bothering with the Agon project or anything else he was doing.

So there you are.

-Gordon

[paulb]

Anyway as it was nearing completion I said I'd not be releasing my code under any open source license but a source available one and that was the last thing I heard from him. Not a peep more. I was a bit miffed especially as I'd put the work into it even bought an expensive programmer which I was unable to use under Linux.

I imagine that if he was wanting to have software to distribute to educators, a Free Software licence would have been preferable, but there would presumably have been room for some dialogue on the matter. Clearly, if one is content to have Acorn's proprietary BASIC in the software stack, the licensing of the other parts starts to seem like less of a general concern. I am reminded of the Maximite's BASIC and the Maximite's creator getting upset that Olimex had forked the code, mostly because he didn't understand what the GPL entails.

At the very least, however, these matters can be discussed and people can work towards a satisfactory solution, whether that involves using something else or not. But I suspect that, as is typical nowadays, the software is being treated like a commodity that people will compete to provide at no cost, all to showcase the wonders of the platform. If someone else shows up eager to impress, expect to be dropped in favour of the enthusiastic newcomer.

The whole project was a bit weird anyway - why bother? Was there really any educational value in this dual but mutually exclusive CPU setup?

It would be interesting to get the perspectives of educators, but they often seem to be the last to be queried when technology is incoming. I once had a comment to that effect from someone about the introduction of the BBC Micro back in the day, although that would count as heresy in some circles with all the legacy-building that has been going on.

Not IMO but there you are. There was also a couple others working on it too - in addition to the Z80 there is an ATmega which does the host and filing system side of things (not a million miles from my own SBCs really) and no-one could agree on some common code for that - the 6502 and Z80 sides required different ways to boot, etc.

I can understand the approach where one models a traditional microcomputer using modern glue logic, whether programmable logic devices or even microcontrollers, as long as one defines the role of these things well enough. No-one should realistically expect people to be unearthing complete chipsets that have not been produced for decades. But a dual-CPU 6502/Z80 architecture seems like it is trying to appeal to everyone when I thought that the project wasn't supposed to be about popular appeal.

I am reminded of a crowdfunding project a while ago where someone was going to fab an ASIC with Z80, 6502 and 68000 cores. Why on Earth would anyone realistically want to do that? I think it was partly motivated in refining a chip design workflow, actually, and the idea was to appeal to the retrocomputing scene. They would have been better off targeting a different niche, and no, it didn't get funded.

[gordonDrogon]

Anyway as it was nearing completion I said I'd not be releasing my code under any open source license but a source available one and that was the last thing I heard from him. Not a peep more. I was a bit miffed especially as I'd put the work into it even bought an expensive programmer which I was unable to use under Linux.

I imagine that if he was wanting to have software to distribute to educators, a Free Software licence would have been preferable, but there would presumably have been room for some dialogue on the matter. Clearly, if one is content to have Acorn's proprietary BASIC in the software stack, the licensing of the other parts starts to seem like less of a general concern. I am reminded of the Maximite's BASIC and the Maximite's creator getting upset that Olimex had forked the code, mostly because he didn't understand what the GPL entails.

Indeed - I'd also gotten Comal, EhBASIC and CBM2 Basic running under it too (Easy to get the Microsoft Basics going as I'd already patched them for my own Acorn-like SBC). I could probably have made other Acorn ROMs work (Maybe Forth or BCPL) if I was bothered to.

And he's using the Z80 BBC Basic on the Agon too.

I felt "Source Available" was good enough. It would not have stopped him selling it.

It would be interesting to get the perspectives of educators, but they often seem to be the last to be queried when technology is incoming. I once had a comment to that effect from someone about the introduction of the BBC Micro back in the day, although that would count as heresy in some circles with all the legacy-building that has been going on.

Today? If it's not BBC Micro:bit then it's not going to happen - at least in the UK. Maybe some of the more well-heeled schools can do the code club thing with a Pi (I did the code club/STEMnet ambassador thing for a short time before lock-down).

I think schools have quite a hard time right now re. computing and so on. Not sure they want to look at anything that's not a little trinket you can plug into a PC running MS Windows..

-Gordon